Solaris 10 to be Released Late in 2004

ptolemu writes "The Register has the scoop on Sun's latest iteration of Solaris. The article includes some details of the new and improved features that will be included in the OS. The OS is scheduled to be released in the second half of 2004."

Comments

so what's better, bsd, linux or solaris?

[Anonymous Coward]

i always wondered why nobody ever writes articles that include solaris.

Re:so what's better, bsd, linux or solaris?

[0xfc]

> i always wondered why nobody ever writes articles that include solaris.

1. not open source
2. costs money
3. runs on overpriced hardware
4. bsd and linux can do everything it can cept maybe scale to extremes
5. solaris is not the only stable OS anymore
6. way too many people were burned by sun back in the day and said enough is enough, they never went back

Re:so what's better, bsd, linux or solaris?

[SpaceCadetTrav]

7. Uses that funny keyboard.

Re:so what's better, bsd, linux or solaris?

[Anonymous Coward]

4. bsd and linux can do everything it can cept maybe scale to extremes

Actually Linux runs a 512 CPU supercomputer at NASA. I think Linux 2.6 is now at least as scalable as Solaris if not more so.

Solaris goes to IIRC 108 CPUs...

IRIX, however, is in a league of its own.

Re:so what's better, bsd, linux or solaris?

[wwwillem]

Linux runs a 512 CPU supercomputer at NASA

Are you talking here 512 CPU SMP or more a Beowulf or similar. Two rather different animals....

Re:so what's better, bsd, linux or solaris?

[BiggerIsBetter]

1. Yup.
2. Yup, but it's cheaper than RedHat/SuSE Linux for a single CPU *professional* version.
3. There's an x86 version (Sun harware fanboys can STFU about how crappy it is on non-SPARC hardware), and low-end Sun hardware starts at around 1000 USD.
4. Yup.
5. Fair enough.

Re:so what's better, bsd, linux or solaris?

[PacoTaco]

Uses funny everything.

Re:so what's better, bsd, linux or solaris?

[SuperBanana]

8.When you plug the keyboard in, the bloody thing crashes!

Re:so what's better, bsd, linux or solaris?

[Anonymous Coward]

1. not open source
2. costs money
3. runs on overpriced hardware
4. bsd and linux can do everything it can cept maybe scale to extremes
5. solaris is not the only stable OS anymore
6. way too many people were burned by sun back in the day and said enough is enough, they never went back

Lets flesh that out a bit...

1. You can get the source to Solaris.
2. You can download Solaris for free.
3. Solaris runs on good hardware which is a good thing if you are trying to get serious work done. (Not everyone working with *nix is building web servers, internet hosting, or using samba to replace a few Windows PCs.) If you are only trying to recycle crap hardware, any OS will do. FreeDOS or DR DOS will recycle hardware that Linux is too fat to run on.
4. BSD and Linux lack the thousands of mature, commerical applications Solaris has, but they are catching up.
5. Solaris is not only stable, it is one of the best. Linux is still in catch up mode in terms of standards and features. Linux still has a tendency to cheat, or only partially implement a standard. It is getting better. Standards are a good thing if you are trying to get equipment from multiple vendors to work together.
6. Sun's support has been plenty good for the companies I've worked for, and PCs won't be getting the work done that we do anytime soon. Maybe if the Opterons work out well we could use them in a couple of years.
7. A standard Sun keyboard has the control key where it should be.
8. Documentation. Solaris has it. The documentation is good, and correct. Linux, ha.
9. Solaris can have a System V Unix personality, a BSD personality, a GNU personality, or traditional Sun personality, depending upon your path.
10. Linux pretty much provides a subset of what Solaris can do.

I could go on, but you should get the point by now.

Re:so what's better, bsd, linux or solaris?

[fferreres]

512 cpus, single image. It's clustered at 64 cpus per node, but they share memory and the same kernel. I am quoting by memory, so may be wrong.

Re:so what's better, bsd, linux or solaris?

[Anonymous Coward]

3. I just got a Sparcstation 20 for FREE! And I have it running Solaris 9 core. I'd argue lowend Suns don't cost 1000's ...at least not from my personal experience :^p

Re:so what's better, bsd, linux or solaris?

[0xfc]

> 1. You can get the source to Solaris.

google showed this link for seach "solaris source". From this link it reads:
The Solaris 8 Foundation Source Program has concluded. Source code for the Solaris Operating System is available for qualified educational institutions and partners; please contact your Sun sales team for details. Maybe this info is old...

> 2. You can download Solaris for free.

I just tried, it seemed you were right, they made me jump through hoops. I ended up at a page that appeared to allow me to download solaris, but the link was not a link and they wanted me to register. could have told me that at the start... i am too lazy to do it, i will assume you are right.

> 3. Solaris runs on good hardware which is a good thing if you are trying to get serious work done. (Not everyone working with *nix is building web servers, internet hosting, or using samba to replace a few Windows PCs.) If you are only trying to recycle crap hardware, any OS will do. FreeDOS or DR DOS will recycle hardware that Linux is too fat to run on.

You make it sound like ibm,hp,compaq did not make high end x86 servers. Himilaya non stop servers come to mind. Heck even proliants are nice.

> 4. BSD and Linux lack the thousands of mature, commerical applications Solaris has, but they are catching up.

I agree.

> 5. Solaris is not only stable, it is one of the best. Linux is still in catch up mode in terms of standards and features. Linux still has a tendency to cheat, or only partially implement a standard. It is getting better. Standards are a good thing if you are trying to get equipment from multiple vendors to work together.

FreeBSD is amazingly stable. Uptimes of a year are taken for granted. I dont know enough to comment on the rest of your statement.

> 6. Sun's support has been plenty good for the companies I've worked for, and PCs won't be getting the work done that we do anytime soon. Maybe if the Opterons work out well we could use them in a couple of years.

Support? I dont need no stinkin support. You telling me your head admins cannot troubleshoot hardware?
You dont have a backup system ready so a hardware failure just is an inconvienence? Software is a whole different issue.

> 7. A standard Sun keyboard has the control key where it should be. ...

> 8. Documentation. Solaris has it. The documentation is good, and correct. Linux, ha.

I agree. Sun probably employs a crap load of technical writers. Its a good thing. I often find answers to problems from sun docs...

> 9. Solaris can have a System V Unix personality, a BSD personality, a GNU personality, or traditional Sun personality, depending upon your path.

I guess.

> 10. Linux pretty much provides a subset of what Solaris can do.

I wont argue that. Solaris is time tested and powerful.

Re:so what's better, bsd, linux or solaris?

[jobsagoodun]

7. You always have to install the GNU tools on it anyway because the built in ones are so shit.

Re:so what's better, bsd, linux or solaris?

[hdparm]

4. BSD and Linux lack the thousands of mature, commerical applications Solaris has, but they are catching up.

Care to explain this bit, please? I am not arguing the number since I don't know Solaris too well but 'thousands' makes me really curious.

Re:so what's better, bsd, linux or solaris?

[Anonymous Coward]

You calling Beowulf an animal now?

Re:so what's better, bsd, linux or solaris?

[0xfc]

I can think of one right now.
navini.com has really nice wireless
euqipment. Their base station requires
either a sun box or a windows box.

what one do you think I will choose?

Re:so what's better, bsd, linux or solaris?

[Anonymous Coward]

The x86 version is now a free download again Sun have waived the $20 download fee. You do have to register first http://survey.sun.com/servlet/viewsflash?cmd=showf orm&pollid=Red%21sol9_x86_download

Re:so what's better, bsd, linux or solaris?

[Anonymous Coward]

No it isn't clustered, although I think SGI's generally available commercial Altix cluster in groups of 64.

No, this is a single system image computer. A single kernel and a single globally addressable memory space.

Re:so what's better, bsd, linux or solaris?

[Anonymous Coward]

SMP

Re:so what's better, bsd, linux or solaris?

[shin0r]

>Support? I dont need no stinkin support. You telling me your head admins cannot troubleshoot hardware?
>You dont have a backup system ready so a hardware failure just is an inconvienence? Software is a whole different issue.

We aren't talking about a few PC's in the basement, or your home ftp server here. For those of us that admin hundreds of machines in production environments, support is absolutely essential - and Sun do it well.

Re:so what's better, bsd, linux or solaris?

[Anonymous Coward]

Yeah well I got a 17" PowerBook for free too. Well... except for the fine I had to pay later and the 12 months of probabation.

Re:so what's better, bsd, linux or solaris?

[Gilesx]

> I could go on, but you should get the point by now.

That you're a Solaris fan-boy?

Re:so what's better, bsd, linux or solaris?

[Mr. Piddle]

Sun released some new UltraSPARC IV Sun Fire servers last week. The top-end one has 144 "cores" (72 physcial CPUs). I'm not sure if they offer single-image clustering, but Sun does offer 8-way traditional clustering with 1GB/sec (8000Gb/sec) interconnect (1152 CPUs and 4TB RAM for appropriate $$$, of course). I've never used one, but can at least drool at their web page.

I'm also suprised that no one seems to be talking about the new 2-way 64-bit Opteron servers Sun is selling, now. If anything should make big talk on Slashdot, this would be it.

Re:so what's better, bsd, linux or solaris?

[Anonymous Coward]

Care to explain this bit, please? I am not arguing the number since I don't know Solaris too well but 'thousands' makes me really curious.

You can access Sun's solutions index here. It might take a little digging to find what you are looking for.

Re:so what's better, bsd, linux or solaris?

[Anonymous Coward]

How funny. See me laughing, funny boy?

Re:so what's better, bsd, linux or solaris?

[Anonymous Coward]

>> 6. Sun's support has been plenty good for the companies I've worked for, and PCs won't be getting the work done that we do anytime soon. Maybe if the Opterons work out well we could use them in a couple of years.

>Support? I dont need no stinkin support. You telling me your head admins cannot troubleshoot hardware?
You dont have a backup system ready so a hardware failure just is an inconvienence? Software is a whole different issue.

Mhmm, yeah right, we all have $2M backup systems sitting around which we can use as replacements. The people who use this high-end solutions from SUN are running serious business applications on them and not simply running a web server or whatever.

I currently administer an E10K loaded up with 64 CPUs, something in the region of 512 Gb RAM (yes Gb)and connectivity to terabytes of SAN storage. When purchased new a couple of years ago it was worth well over $2M.

No, we DO NOT have a spare one tucked under my desk to use when the first one breaks.

Remember that this is serious high-end gear, not web server fanboy PCs.

Re:so what's better, bsd, linux or solaris?

[DrDebug]

>FreeBSD is amazingly stable. Uptimes of a year are taken for granted.

People claiming uptimes of a year or more make me cringe. Don't they even CONSIDER applying the latest SECURITY PATCHES?

Re:so what's better, bsd, linux or solaris?

[Catharz]

You can download Solaris for free.

You can also get it on CD for the cost of postage. I purchased an old Sun UltraSparc some years ago. It cost me half a dozen VB cans and the cost of a new AU$40 clock chip (they have a battery on the clock chip and the machine won't keep time once the battery is dead) and I purchased Solaris 7 for US$10 (cost of postage + CD's) which I thought was a good deal. I was intending to run Oracle on it (for developing against it at home), but it's been a door stop since Oracle was ported to Linux.

Even though the machine I have is ultra slow, it is also ultra stable when compared to my Linux server. But that may be due to the difference in load each machine gets, and of course YMMV.

A friend has expressed interest in using it as an X-Term (it has one of the nice 17" trinitron screens). I'll make him buy me a bottle of scotch (and make a profit :D).

Re:so what's better, bsd, linux or solaris?

[wwwillem]

The large Sun servers (in the past max 72/104 CPUs, now with upto 144/208 cores), are indeed SMP single image. Although most of our (oops, yes I work for Sun :-) customers configure them in multiple smaller domains.

And yes, those Opteron servers are definitely cool. Mmm, I have to take a peek at the current employee discount levels....

Re:so what's better, bsd, linux or solaris?

[tigga]

>FreeBSD is amazingly stable. Uptimes of a year are taken for granted.

People claiming uptimes of a year or more make me cringe. Don't they even CONSIDER applying the latest SECURITY PATCHES?

It depends... It does not have to be rebooted because of userland vulnerability. You pretty much may ignore local kernel exploits if there are no local users (except you ;). And remote exploits happen not so often lately.

SO?????

[Anonymous Coward]

Will it be called SunOS 2.10 or SunOS 3.0?

Re:SO?????

[MrPerfekt]

Will it be called SunOS 2.10 or SunOS 3.0?

It's actually SunOS 5.10. (SunOS 4.x was Solaris 1.x, SunOS 5.x was Solaris 2.x up until 2.7, then they changed it to just Solaris 7 with the underworkings of SunOS 5.7... got that?)I can't imagine they're going to break into the next major version number. (i.e. SunOS 6) but you never know.

Re:SO?????

[cujo_1111]

So 1 = 4?
And 2 = 5 if 2 And 2 = 7 if 2 > 2.7?

I get it now...

# Segmentation fault (core dumped)

Re:SO?????

[g2racer]

Wasn't the reason they went from v4 to v5 because they swapped the underpinnings of the OS from BSD (Solaris 1.x) to SVR4 (Solaris 2+)? That being said, I can only see them going to v6 when they change over to Linux ;)

Re:SO?????

[DarthBart]

Solaris = SunOS + OpenWindows.

Solaris 7 = SunOS 2.7 + OpenWindows 1.something

Re:SO?????

[DebianRcksLindowsLie]

It's actually SunOS 5.10. (SunOS 4.x was Solaris 1.x, SunOS 5.x was Solaris 2.x up until 2.7, then they changed it to just Solaris 7 with the underworkings of SunOS 5.7... got that?)I can't imagine they're going to break into the next major version number. (i.e. SunOS 6) but you never know.

Geek license awarded! I don't even want to know HOW you know this, much less keep track!

--
The best review of Lindows ever published is in my sig!

Re:SO?????

[Anonymous Coward]

Well, actually, Solaris 1.x is a bit of revisionist history. It was always just referred to as SunOS 4 until after the release of Solaris 2.0. Then, when they had to issue a new version of 4.1.3, they branded it as Solaris 1. So, Solaris 2 predated Solaris 1.

Re:SO?????

[Anonymous Coward]

Wrong that is not a core dumped, is the logic when you go to pay taxes :)

Re:SO?????

[Anonymous Coward]

Well, at least I got whoever M'd this one with an "unfair" in M2.

Sub roots

[AmVidia HQ]

Sun has also added a new security tool with Solaris Privileges. This lets the root user create sub roots that can have permission, for example, to patch applications but not to touch hardware components.

When will I see it in Debian stable? =b

Re:Sub roots

[theantipode]

Considering Debian stable, last I checked, still has the 2.2 kernel as default, I'd say you have at least a ten year wait.

Re:sub roots

[Russ Steffen]

This feature sounds like the privilege model from Trusted Solaris is being mainlined into the plain ol' Solaris tree. In which case, yes, someone is working to bring that into Linux. That's one of things SELinux is doing.

Re:Sub roots

[KrispyKringle]

Sounds vaguely similar to sudo.

Re:sub roots

[Anonymous Coward]

The kernel can do it no problem. Userspace will cotton on sooner or later.

Re:Sub roots

[Frymaster]

Sounds vaguely similar to sudo

you can easily roll your own one of these with a combo of sudo and acls.

but of course if you let sun do it for you a) you save yourself some work b) management feels more comfortable about it.

Re:sub roots

[Anonymous Coward]

In linux you can set up SELinux.

this is Security Enhanced Linux.

It basicly isolates every thing from everything else in linux right down to the kernel level.

For example if you have a Apache webserver and it gets comprimised, a hacker can't use Apache's security level to give him elevated permissions to control another part of the OS. In a regular OS you have to allow the Apache some root control over the computer to have it work properly and a hacker can use this to violate your computer.

In SELinux even if a hacker gained root access their is a limited amount of damage he can do, depending on how you set it up.

You could if you wanted to use this to set up roles for users, like a apache admin or a sendmail admin, or a filesystem admin or a /dev/ file admin.

SeLinux is brought to us by our freinds and future government overloads: the NSA.

Re:Sub roots

[Anonymous Coward]

Check your facts troll

Re:sub roots

[segment]

actually using something as simple as groups can achieve this in case you (obviously) didn't know. ACL's work fine and lsacl written in Ruby can be tweaked to run elsewhere. But just in case you don't feel like tweaking, or perhaps you just ... don't know how... ACL Linux can be downloaded. Personally I prefer solaris over most flavors of *nix, and I've been through them all starting about 6 years ago with Linux. As an admin, for financial purposes... I would go with the following in this order... Sun* (including trusted sol), AIX, HPUX, BSD, Linux

Re:Sub roots

[LinuxGeek]

Didn't he play the Gimp in Pulp Fiction?

Re:sub roots

[ryanw]

Interesting... Everyone has the debate of Linux GPL vs putting Linux out there with a BSD license. Would would be the problem of letting an older linux kernel tree off and call it something else and go with a BSD license? If it dies a horrible death, so be it, the main GPL linux tree will continue... But that way it could give the debate a run for it's money.

Re:sub roots

[Anonymous Coward]

Yeah wooh! Then Microsoft could pick it up and start a monopoly... hey wait a sec-

Re:Sub roots

[Anonymous Coward]

It's been there for years with the various ACL patches... Google is your friend, search for "debian acl"

Re:Sub roots

[Anonymous Coward]

And don't forget c) it won't run on a useful OS anymore.

Re:sub roots

[ajagci]

Yes, you can get similar functionality for Linux.

From a security standpoint, it's a bad idea. But it markets itself well because everybody thinks that more security features mean more security.

Re:sub roots

[0xfc]

> For example if you have a Apache webserver and it gets comprimised, a hacker can't use Apache's security level to give him elevated permissions to control another part of the OS. In a regular OS you have to allow the Apache some root control over the computer to have it work properly and a hacker can use this to violate your computer.

If one hacks apache remotely the permits you will gain is uid nobody. I am not quite sure how you leaped to uid 0 access from that alone. No one i knows run apache as root except irix admins. :P

> In SELinux even if a hacker gained root access their is a limited amount of damage he can do, depending on how you set it up.

Reminds me of FreeBSD jails I have been using. FreeBSD's killer app, one of many.

> You could if you wanted to use this to set up roles for users, like a apache admin or a sendmail admin, or a filesystem admin or a /dev/ file admin.

Last time i checked in the /etc/passwd file, we have users such as bin, operator, smmsp, man, etc... Some of them are used, some have gone the way of a green pasture.. but just because they are out of style with some people does not mean they never existed. I know you get fine grained controls from SEL, but one can do a lot without it.

> SeLinux is brought to us by our freinds and future government overloads: the NSA.

**** them and use trustedbsd. their mascot is cooler anyways.
http://www.trustedbsd.org/beastie.html

Re:Sub roots

[bruthasj]

It already is!

Re:Sub roots

[Imperator]

It would be more than that, at least from what the description suggests. The problem with sudo is that you're often giving suid access to programs that aren't designed to be suid, so someone who was the right entries in the sudoers file can root the machine with ease. Proper privilege separation in the admin tools would mean being able to give someone access to run apt-get dist-upgrade (or whatever it is) without his being able to install his own packages. It would mean letting someone add non-root users but not root users, or resetting passwords but only for users in a certain group. It requires planning when creating admin tools, not a "slap it on" solution like sudo.

Of course, given that it's Solaris, it may end up just being sudo after all.

Re:sub roots

[Imperator]

In a regular OS you have to allow the Apache some root control over the computer to have it work properly and a hacker can use this to violate your computer.

No, it just has to start as root if you want it to bind to a privileged TCP port like 80. Very few servers require root access once they bind. Apache, like many other daemons, can be run as an unprivileged user in a restricted chroot jail.

Re:Sub roots

[axxackall]

Since when management feels more comfortable in situations where sysadmin feels it less? Oh, and did I tell you that sysadmins do not feel comfortable when Sun attempts to do any job for them screwing something up instead?

Re:Sub roots

[Anonymous Coward]

Shhh! I think he's sleeping...

Re:sub roots

[damiam]

That'd be impossible. Every person who had ever contributed any code to the kernel would have to agree to relicense their code. Mozilla tried this a while ago and had a bitch of a time tracking everyone down, and that was for a project just a couple years old.

Re:Sub roots

[SW6]

> Sun has also added a new security tool with Solaris Privileges. This lets the root user create sub roots that can have permission, for example, to patch applications but not to touch hardware components.

When will I see it in Debian stable? =b

I already have it on my woody boxes, thanks to Linux-Vserver. And it works a treat.

Re:Sub roots

[armb]

> Sounds vaguely similar to sudo.

Maybe more like userv
http://www.chiark.greenend.org.uk/~ian/user v/

Re:Sub roots

[Anonymous Coward]

you can easily roll your own one of these with a combo of sudo and acls

Actually, no you can't. There's nothing easy about it. RBAC (Role Based Access Control), which is an advancement on sudo, has been in Solaris since Solaris 8. Least privilege goes even further.

With RBAC/least privilege you can control things down to not just commands and options, but even down to specific pieces of hardware. Also, since an account has individual privileges, you could in essence take away privileges from root (consider a root account that couldn't su to certain users or read those users files, but still could su to anyone else and read any other files).

Re:Sub roots

[segfaultcoredump]

Not quite.

One simple example of how they differ would be in the running of a process such as apache that needs to bind to a "privileged" port.

Lets say you wanted to allow a regular user to run apache. Normally they need root (or a suid script that ran apache for them) so that it can bind to port 80. However you dont want to give their app (apache) privs to read/write every file on the system or do any of the other things that root can do.

with priv sep, you can grant that user the ability to bind to any port, but no other privs. With sudo, it is all or nothing.

it is a subtle, but rather important, difference.

Another difference is how things get audited. With priv sep, the audit daemon records the fact that I did something, not root. Thus you dont have to dig through the logs to try and figure out who sudo'ed to root and what they did once they where there

Re:Sub roots

[nslu]

well. in sudousers you define command _with_ parameters.
so you can limit user to running "apt-get dist-upgrade" only, not any other apt-get action.

Re:sub roots

[Anonymous Coward]

> In SELinux even if a hacker gained root access their is a limited amount of damage he can do, depending on how you set it up.

Reminds me of FreeBSD jails I have been using. FreeBSD's killer app, one of many.

It is not jails. It is domain and type enforcement. I strongly recommend you read up on this topic before you say "it sounds like like an apple except with bumpy skin, a different color, and totally different internally, but hey it's all fruit".

Re:Sub roots

[AmVidia HQ]

thanks, I had trouble with CHROOT and this would seem like a very nice way to do virtual serving.

N1 Grid Containers Look Interesting

[TravisWatkins]

It should be interesting to see how the N1 Grid Containers work. It would be great to setup a shared server with this so scripts can't eat all the CPU and crash the entire server.

Re:N1 Grid Containers Look Interesting

[SpamSlapper]

If N1 Grid Containers work well it will be a major improvement on the competition. In HP-UX if you want to set up V-Pars you need to dedicate at least one CPU, physical disks and a network interface to every partition, and resource allocation is at the whole CPU level. With an 8 CPU machine that doesn't give you much leeway if you want to have 3 or 4 test environments.

With N1 Grid Containers OS instances sit on top of a "master" OS, so resources can be divided at a much finer level. You could presumably have a production partition with 80% of the cpu power allocated to it, and a bunch of test partitions sharing the rest, and dynamically increase the CPU power of the test partitions when it was prudent to do so.

Re:N1 Grid Containers Look Interesting

[Imperator]

Posix has had resource limits for a long time now. See setrlimit(2), for example. PAM has had a module to support this for quite a while. I'm sure Solaris has resource limits. Virtualization (what the N1 system seems to be) is a whole different beast, with different uses. If all you want is to stop a script from eating all the CPU, any Unix will do that for you.

Re:N1 Grid Containers Look Interesting

[jsavit]

The resource management part has been in Solaris for quite a while, letting you control how much CPU (and other resources) an application can get. What the containers add is the ability to provide isolated environments (namespace, filesystems, process list, security and fault containment contexts) so each container thinks it has its own instance of Solaris.

Re:N1 Grid Containers Look Interesting

[christophersaul]

Solaris Resource Manager currently lets you have similar functionality in software, with dynamic system domains take it further, allowing you to make changes to hardware configs on the fly whilst the system is running, unlike with HP, which requires a reboot.

Is Unix Unix?

[ObviousGuy]

I've wondered for a while now, is one Unix like another Unix? I've used Linux in the past and am trying out FreeBSD now. Frankly, I don't notice the difference from an end-user perspective.

Linux has SMP support, so does FreeBSD, and so does Solaris. They all have process management functionality (which is what Solaris is introducing with N1 Containers in this release). What would possess me to use Solaris (which costs) instead of Linux or FreeBSD (which are free)?

Is any one of them more robust than another?

Re:Is Unix Unix?

[Anonymous Coward]

Basically it's about filesystem layout and scripts, and to some extent commands; NetBSD has the most sane layout and scripts out there, followed by FreeBSD, IMNSHO.

So, it's all about sane system design. For an inexperienced user the differences are neglectable, but for power users the BSD's really shine.

Re:Is Unix Unix?

[Anonymous Coward]

IMNSHO

... err, OK then if that is your not so honest opinion, then what is your honest opinion?

Re:Is Unix Unix?

[Anonymous Coward]

I think he means humble not honest.

Re:Is Unix Unix?

[Anonymous Coward]

What are you daft? Without even getting in to the many companies, like IBM, that hire people to work on OSS projects, let me just state the provably obvious: Closed source companies are outsourcing their programming jobs overseas. You really need to pick a better place to spread such FUD. I'd recommend MSN or AOL.

Re:Is Unix Unix?

[Anonymous Coward]

If you care at all about programmer's jobs, you would support closed-source Unix. This open source bullshit is causing the current IT job downturn. I don't care if it's enterprise Unix or Windows, just stay away from Linux and *BSD if you want programmers to be employed.


Hello again Darl.

Re:Is Unix Unix?

[Peter Cooper]

The H in IMHO and IMNSHO stands for "humble," not "honest."

Re:Is Unix Unix?

[Frymaster]

What would possess me to use Solaris

one word: support.

i have worked in two shops in the last four years. one is a red hat shop. we use rhel es with paid support. the other was a full-meal-deal sparc/solaris shop.

in the solaris shop we had a dramatic failure of a storedge sena array. i called the sun support line and a guy in tweed jacket was at my door in 40 minutes with a grocery bag full of spare parts (gbic cards, if you care). the problem was solved in a total time of one hour.

in the linux shop i made a web support request for a very simple question (that being: is stronghold bundled with rhel es like the marketing material says? it doesn't seem to be... anyone know?). i logged that request twelve days ago and it's still listed as "awaiting technician". twelve days! and every time i go to check the status the web page throws a NullPointerException. and i got an email for resolution on a support request i didn't even make. i informed red hat that i'd received someone elses support mail and they replied that it would be rerouted, but the erroneous issue still shows up on my incident tracker a week later.

so... sun costs a bundle. but if you need tech support from a team that makes the justice league of america look like a quilting bee, they're your guys.

Re:Is Unix Unix?

[deek]

• I've wondered for a while now, is one Unix like another Unix? I've used Linux in the past and am trying out FreeBSD now. Frankly, I don't notice the difference from an end-user perspective.
  

Well, from a basic end-users perspective, there isn't much different at all. Especially if you install a bash shell on solaris, or whatever unix you're using.

From the administrators perspective, there can be a world of difference. Many admin tasks can be very similar, but many are also pretty different.

As for why you'd want to use Solaris over Linux, nobody does NFS better than Sun. I'm not sure what the current status of NFS is under Linux, but I've heard some stories that don't look favourably on Linux and NFS.

Also, Solaris performs and scales very well on multi-cpu machines, compared to Linux (although with Linux 2.6, this may not be such an advantage anymore).

Then you've got the added advantages of Solaris being a full 64 bit OS (ignoring the Intel version), with large max file sizes and RAM without any special hacks (again, Linux 2.6 has gone some way to fixing this, with 64 bit file support).

Basically, for the enterprise, Linux wasn't really an option until 2.6. With 2.6 only in its infancy, Linux still isn't an option. Solaris is though, 'cause it's got the features, the performance, the reliability, and it has been thoroughly tested on the anvil of time.

Re:Is Unix Unix?

[0xfc]

I will bite.

> If you care at all about programmer's jobs, you would support closed-source Unix.

This does not seem very logical. It almost seems like you are "scared" of OSS. You, as a programmer can now harness all this free software to create some amazing products. With skill you can add value and sell it. Imagine working from scratch on everything, reinventing the wheel at every step, buggy beta code OR you can build on a well written, free, BSD licensed piece of software AND not share your source while making money. You can be even more revolutionary and picking GPL licensed software and sharing your value added source with everyone while making money (support + modest price of software).

I am more a customer, and I cant imagine not having source to compile from now days. I am way to used to it, it is assumed.

Re:Is Unix Unix?

[Anonymous Coward]

I'm not sure what the current status of NFS is under Linux

It works very, very well. We mix Linux/x86 NFS servers with Solaris/SPARC ones, and they're all solid as a rock.

If there's any stability or performance advantage to Sun's NFS implementation, it's negligible. I can't remember the last time we lost an NFS server.

Re:Is Unix Unix?

[This is outrageous!]

i called the sun support line and a guy in tweed jacket was at my door in 40 minutes with a grocery bag full of spare parts

Oh my, what /. is turning into.

Re:Is Unix Unix?

[whereiswaldo]

in the solaris shop we had a dramatic failure of a storedge sena array. i called the sun support line and a guy in tweed jacket was at my door in 40 minutes ...

in the linux shop i made a web support request for a very simple question ... i logged that request twelve days ago and it's still listed as "awaiting technician".

You must admit that these two issues are *very* different in severity. Try logging a failure of a similar magnitude with Red Hat and report your results back for an apples to apples comparison.

Re:Is Unix Unix?

[Anonymous Coward]

Your post basically sounds like exactly what someone who doesn't know what they're talking about would say.

You don't write for ZDNet, do you?

Re:Is Unix Unix?

[Veridium]

I have to agree with you. Sun support kicks butt, second to none. At the one company I worked at as a Solaris admin, the few times I did call Sun, I was never on hold more than a few minutes, and whatever parts were needed were delivered to us within hours.

Though I always saw that as one of the advantages of having the OS & hardware coming from the same vendor. It seems to keep them from playing the "it must be your hardware" game that so many software vendor support people play whenever the answer isn't easy. Though that doesn't explain your experience with RH.

Anyone out there have experience with their X86 support?

Re:Is Unix Unix?

[Anonymous Coward]

UNIX is like sex, it's better when you don't have to fake it.

Re:Is Unix Unix?

[ajagci]

in the solaris shop we had a dramatic failure of a storedge sena array. i called the sun support line and a guy in tweed jacket was at my door in 40 minutes with a grocery bag full of spare parts (gbic cards, if you care). the problem was solved in a total time of one hour.

And if you had bought Linux machines for the same amount of money, you could have bought so many extra machines that you could have just pulled that machine, dumped it in the trash, and gone on merrily with your work.

in the linux shop i made a web support request for a very simple question (that being: is stronghold bundled with rhel es like the marketing material says? it doesn't seem to be... anyone know?).

Well, how much money are you paying RedHat? Anywhere near what you are paying Sun? Even within two decimal orders of magnitude? I suspect not. For what you are paying Sun, you could get stellar Linux support if you wanted to. But somehow, you have concluded that it's OK to pay a lot of money to Sun for service, but that it's not OK to pay a lot of money for Linux service.

Also consider yourself lucky that you have not experienced Sun at their worst yet: yes, they, too, sometimes have less than stellar service even though you pay them a bundle.

Re:Is Unix Unix?

[stevens]

Also consider yourself lucky that you have not experienced Sun at their worst yet: yes, they, too, sometimes have less than stellar service even though you pay them a bundle.

Like diagnosing a memory card failure, replacing it to have the OS panic in a few hours? And then they changed out all the RAM again. And did it again three days later.

After a few calls the replaced everything but the chassis. But that was several failures in a week!

Or when our CPUs started blowing, but they wouldn't give a new batch for all the identical severs we bought. They told us to call them as the blew. We ended up getting about 60% replaced, as they blew and caused downtime.

But they're always there in an hour to be unhelpful.

Re:Is Unix Unix?

[Anonymous Coward]

Yeah because a couple of shitty x86 machines can do the job of a SAN just as well can't it?

Oh and saying "I suspect" instead of actually taking the time to check your damn facts just makes you look like an even bigger idiot, should that be possible.

Now get back to us in five years, at least once you've graduated from High School.

Re:Is Unix Unix?

[kshcsuf]

Some of the advantages over the free UNIX variants that Solaris possesses include things like Dynamic Reconfiguration (did your system board or CPU or physical memory or network interface or SCSI/fiberchannel interface just fail? That's okay, just cfgadm -c disconnect it and swap it out), mpxio (OS-based multipathing of fiberchannel I/O), hotplug support (mostly falls under dynamic reconfiguration), ipmp (IP multipathing -- dynamic failover of network interfaces). This functionality is built into the operating system, and most of it works on even the most basic Sun-based hardware. Things like system board-level DR require workgroup servers or better.

Re:Is Unix Unix?

[silence535]

one word: support.

Hmm, I vaguely remember ideling around for a couple of days with a bunch of other expensive developers because Sun Gold(!) support was not able to quickly fix a hardware failure.

And worse, it turned out that the failure was due to a combination of a hardware raid controller and a mainboard that was known to fail under circumstances and was forbidden for (from|to) sale by a Sun directive and no one knew why the f*ck they had sold us this combo.

And this fine machine came with a big bucks (tm) 3d OpenGL accelerator card and a 21" Monitor to sit in the server cabinet and watch the console on. Stupid. But then again this might have been the mistake of our materials buying department.

-silence

Re:Is Unix Unix?

[bruthasj]

I run NFS in Solaris clusters and in Linux clusters. Both have issues.

Solaris NFS doesn't scale very well.

Coupled with flaky Solaris NIS, NFS has some nightmarish support scenarios.

Linux NFS doesn't lock very well. (rpc.statd crashes ALL the time)

Slightly different NFS versions (Rh7.3 v. Rh9) can crash rpc.mountd as well on Linux.

Re:Is Unix Unix?

[Anonymous Coward]

You must admit that these two issues are *very* different in severity. Try logging a failure of a similar magnitude with Red Hat and report your results back for an apples to apples comparison.

How you do anything is how you do everything.

Re:Is Unix Unix?

[Anonymous Coward]

Though I always saw that as one of the advantages of having the OS & hardware coming from the same vendor.

Then why does mac support suck? Every time i get something back from them, something new is broken.

Anyone out there have experience with their X86 support?

Dell usually gets us replacement equipment w/o question within 1-2 hours under our 4 hour service, so that's good compared to Sun's 3-4 hour response underneath a simlar plan. But of course YMMV depending on where the local support centers/storage areas are. Once you can place a call saying "this is what's broken" as opposed to just a "help", then, in my experience, most service centers will follow your lead.

Re:Is Unix Unix?

[Anonymous Coward]

flacky Solaris NIS ?

I ran NIS on a 3000 server Solaris environment for nearly 10 years with no NIS failures outside of 3 human errors in that entire time. Only now are we dumping NIS in favor of secure LDAP.

NFS has always worked well in our high throughput compute environment as well. Automounter under Solaris works as well...took them a few years, but look at Linux it still doesn't work right and the code is several years untouched for most major issues.

Re:Is Unix Unix?

[Anonymous Coward]

I have to agree. Support is the big difference. We are migrating some of our Sparcs running Solaris 7 to Dell 2650s running RH Linux ES. Whats funny is where I putting these was in a large govt contractor whose mainly a windows shop at the branch I am located at.

Suns support was amazing, if we had any problem I either had someone on the phone in under an hour or onsite in under 6 hours.

With RedHat I get better support using the newsgroups. RH support is a big joke. When we got the 2 systems in the lab and loaded them up, the first problem we had was getting rhn_register to work (i had already updated the rpms for the new keys). A day passed as RH could not find our product ID and referred us to Dell. So I call Dell, I get support immediately and they conference call in RH9 Support.

Get this the cluemonkey at RH said I need to downgrade the kernel to make it work....huh? I cant even upgrade it yet.

Anyhow while Dell is trying to get RH to answer I find the solution in an old RH mailling list post (rhn_register wont work if theres no host entry for the system and no forward dns). I added the hostname and ip in /etc/hosts and rhn_register worked.

New Problem the Product ID key doesnt work. Guess what it has been A WEEK and Dell finally gave up with RH and is sending us (at their cost I would guess) a new product key SO WE CAN PATH RH ES shit.

I loved RH until I bought their enterprise line. Heck I still use Fedora at home and get better mailling list support.

RH Enterprise my ass.

Re:Is Unix Unix?

[maitas]

Sorry, Solaris is free... go to www.sun.com/downloads/ and get it! You can download SPARC or Intel versions.
Acording to my own experience, JBOS on Solaris Intel is 10% faster than on Linux, mainly becouse of the better thread management Solaris has (I haven't try kernel 2.6, thoug).

Re:Is Unix Unix?

[TheLink]

Some customers spec big screens for their servers, because they are going to use them elsewhere and it's easier to get them that way.

Re:Is Unix Unix?

[TheLink]

Nonono, you'd support Windows. There's always PLENTY of work to go around if you use windows.

I personally like OSS because 1) life is too short to keep rewriting code, 2) it's free - life is too short to have to keep saving up to not do everything yourself. Why rewrite code that a whole bunch of people have spent many many years of their life writing and have managed to do a decent job of it. The worst is having to rewrite your OWN code, just coz the license belongs to someone else.

Might as well spend years of your life doing something that hasn't already been done 10 times already.

Sure GNU/Linux is basically a reimplementation of Unix, however the difference is, if this OSS stuff continues, you don't have to do all that shit again.

People can keep moving forwards without having to ask for permission/licenses from hundreds or thousands of other people at each step.

Similarly patents don't scale well either.

Re:Is Unix Unix?

[TheLink]

He'd sounds like he'd prefer a job similar to painting the Golden Gate bridge.

Or cleaning/fixing windows. There's always plenty of work, coz windows always get dirty or broken.

Re:Is Unix Unix?

[Servo]

And if you had bought Linux machines for the same amount of money, you could have bought so many extra machines that you could have just pulled that machine, dumped it in the trash, and gone on merrily with your work.


You obviously aren't used to working in a mission critical environment where downtime costs money. It isn't about the hardware, its about the service that system provides. You can't just throw out a production server.

The only place that makes any sense at all is in a dumby web farm where they are all serving the exact content. But since most companies aren't dot coms, they need their systems to keep running.

Re:Is Unix Unix?

[__past__]

This is true today, but I think it will become less so in the future.

Using clusters of redundant, inexpensive servers has some really nice advantages (for example, not only can you "hot swap" complete servers in case of failure without downtime, it is also easier to throw some new boxes in if you need more power than it would be to upgrade a single big-ass server). The problem is that it is hard to design apps to support such a setting nicely, but I predict that more and more will be extended to support clustering (or replaced with alternatives that do).

Of course, that does not mean that big iron unix servers will completely go away, just like the mainframe refuses to die.

Re:Is Unix Unix?

[christophersaul]

Sun's actually a lot, lot less expensive now. Check out the V440, which can come in much cheaper than an equivalent 4 way Xeon.

Re:Is Unix Unix?

[christophersaul]

That's right. Because Linux has all those commercial apps that run on Solaris and if you have a problem with a storage array, using Linux servers is somehow going to solve the problem.

Show me a Linux vendor who can provide anything like the application support Solaris has, along with Sun's platinum support, globally, for both your hardware and OS, with surround products. Apart from Sun, that is.

Re:Is Unix Unix?

[christophersaul]

Predictability - nice quarterly releases, where things just work. A huge suite of commercial applications to choose from on Sparc, loads of hardware support, scalability from one sub $1,000 Sparc or X86 box right up to massive campus clusters and terabytes of storage, one company to call for OS, surround apps (cluster, volume manager, etc) and hardware. In terms of SMP support, Solaris scales much higher than Linux, although few customers will need that. Large telcos, service providers and the like will, however. Excellent performance too, particularly on heavily loaded database driven apps. The new benchmarks for UltraSparc IV show significant improvements over what's available now.

The Sparc boxes can even come in at less than the Xeon equivalent now.

Re:Is Unix Unix?

[christophersaul]

Why are you migrating exactly?

Re:Is Unix Unix?

[__past__]

It is not as free as one might think.

First of all, the free binary license is only available for hardware with a capacity of at most one CPU (no matter how many it actually has), and has been bought from Sun or an authorized reseller (not eBay or something).

Second, you only get a 60-day trial, or a "developer" resp. "educational" license that allows you to use Solaris "for the sole purpose of designing, developing and testing your applets and applications to run on the Solaris platform". I.e. no legal web browsing or using StarOffice for you.

In other words, you cannot generally buy a cheap Ultra 5 from eBay, download a new Solaris version for free, and use it as a private workstation or home server. At least not legally, that is.

(That applies to the Sparc version. The terms for x86 tend to be slightly different (no "authorized reseller" clause, for example), but then again, they tend to change rather frequently.)

Re:Is Unix Unix?

[v01d]

Try logging a failure of a similar magnitude with Red Hat and report your results back for an apples to apples comparison.

My company has a active cluster running RH AS2.1. We pay for their best support contract. Getting RedHat to do anything is difficult and time consuming. I've called for support and been transfered to voice mail with a full mailbox on multiple occasions. I can't call RedHat and get an answer about their offerings.

I can personally call Sun (no business relationship) and get questions answered for any of their products.

Re:Is Unix Unix?

[frission]

also, with as much as the company probably paid for the sparc machines and solaris and support, you could have had a couple of backups or redundancy with the linux machines :)

Re:Is Unix Unix?

[__past__]

You don't have to start from scratch even if you don't use free software, you know? You can just buy a license for proprietary libraries. If this seems unbelievable to you, just ask the sales departments of Trolltech or MySQL AB, they'll happily explain.

I agree however that having access to the source is certainly valuable from a customer point of view, and would often justify a higher price. (The market seems to agree, just look at how expensive source access for proprietary software tends to be.) Then again, for me it doesn't always have to be free in the FSF, or open source in the OSI sense - for example, the right to distribute derived works is significantly less important (although nice to have) than the right to use a patched version internally, or just to be allowed to look at the source to find workarounds when something goes wrong.

Morale: It is not a very sensible position to think that OSS=good and proprietary=bad. It always depends on your specific needs.

Re:Is Unix Unix?

[borzwazie]

One word:

truss.

There is not one single other utility that compares to truss on Solaris for troubleshooting whatever problem you might have.

Re:Is Unix Unix?

[Mr. Piddle]

Is any one of them more robust than another?

It generally depends on what you're doing, what bugs you touch on, etc.

Linux is the all-purpose "Windows" of the UNIX world. No offense, but this is really becoming more true every day.

The BSDs vary, but are generally world-renowned for network infrastructure. I never did figure out Linux firewalling, for example, but OpenBSD was a piece of cake by comparison. The BSDs tend to be more intuitive, better documented, and more coherent than most Linux distributions (not all, of course).

Solaris is a world-class workstation and server OS. It is commercial, so you don't get all the source code. It is, however, very well documented, and Sun generally doesn't keep too many aces up their sleeves (the kernel is documented in a book called "Solaris Internals" and it is UNIX plus X11R6, for the most part). Solaris + Sun hardware is a pretty sweet combination (analogous to Mac OS X on Macs, I suppose), where nearly everything does "just work", if you don't mind reading a bit, first.

Re:Is Unix Unix?

[GAVollink]

I purchase machines in this way myself. This years bleeding edge back-end server is a really great desktop 5 years from now... but only if you've got a decent frame-buffer driver. I have a SPARCServer 1000 in my basement - as a firewall, but if it had a decent frame-buffer, I'd use it as a desktop. Not willing to make that mistake again. Servers get decent video cards - so that when I buy them (or sell them) in a few years they will actually be worth something.

Re:Is Unix Unix?

[jazzmanjac]

Basically, for the enterprise, Linux wasn't really an option until 2.6. With 2.6 only in its infancy, Linux still isn't an option. Solaris is though, 'cause it's got the features, the performance, the reliability, and it has been thoroughly tested on the anvil of time.

Linux was good good enough for Amazon.com back in 2001. They replaced SUN and HP with linux.

Re:Is Unix Unix?

[whereiswaldo]

How you do anything is how you do everything

Do you wipe your ass as fervently as you make love to your wife?

Re:Is Unix Unix?

[Anonymous Coward]

From a big Aus university point of view, Sun used to be great but now they've really dropped the ball. They take a day or so to pitch up now (which is still better than some) and have changed their support to 4 hour "call back" not 4 hour "fix". And they tend to not have parts in stock that you'd think would be fairly basic - like mobos for Ultra 10s or even RAM?? (They insist you use their RAM only...)

You know who provides seriously superior support to us atm? Dell... they really go out of their way for us - but I think we're still in the honeymoon phase, so time will tell how long this lasts.

Re:Is Unix Unix?

[Anonymous Coward]

And if you had bought Linux machines for the same amount of money, you could have bought so many extra machines that you could have just pulled that machine, dumped it in the trash, and gone on merrily with your work.

No really. What level do you want to do a comparison on, and do you want to compare no-name, white box systems with no enterprise features or would you like to use something reasonable like HP ProLiant DL servers? If you take the support costs out of the mix, a similarly configured DL380 and a V240 is about a 25% cost differential (about 2950 versus 4200). So, how many extra machines can you buy with a 25% cost differential? About 1/4 of a machine.

Well, how much money are you paying RedHat? Anywhere near what you are paying Sun? Even within two decimal orders of magnitude?

Again, if you want to compare apples to apples. HP 3y 24x7x365 4h support for the hardware is $949, plus $4997 to RH for 3 yr 24x7x365 4h support for RHAS 3.0 standard support subscription (total $5946). Or, I can get 3y 24x7x365 4h Gold+ support from Sun for $4400 for both HW and software.

So, let's see... 2950+5950 = 8900. 4200+4400 = 8600. Wow, my Sun is 300 cheaper.

Re:Is Unix Unix?

[Anonymous Coward]

Uh wtf,

How about 'strace', 'ktrace', 'ptrace' et al? Same damn thing, different OS.

Thanks, come again.

Re:Is Unix Unix?

[Anonymous Coward]

Yes, Red Hat ``PHONE'' support rocks, their web support was horrible (2001-2003), it may be better now.

I worked in a shop that had `premium' Red Hat support. I called them several times for driver support for arbitrary hardware, kernel modules for clustering...etc. The technicians who answer their premium-support-lines know Linux well. You didn't get passed to five or more technicians before you got your answer. I was pleased with the skill and resourcefulness of RedHat's premium-phone-support group.

Re:Is Unix Unix?

[scrytch]

I've wondered for a while now, is one Unix like another Unix? I've used Linux in the past and am trying out FreeBSD now. Frankly, I don't notice the difference from an end-user perspective.

Basically, there isn't, especially between Linux and FreeBSD. If you were working on a commercial unix system a few years back, you'd be using CDE, which does feel different (usually clunkier, sometimes slicker, e.g. drag and drop onto the panel calendar to accept a meeting). It certainly looks different: it makes windows 3.1 look downright pretty. Nowadays, Solaris still has CDE but is pushing toward GNOME, and other unix vendors are also taking in KDE or GNOME. CDE will still be around for a while yet, but it's certainly no longer the popular choice for developers or users.

Linux has SMP support, so does FreeBSD, and so does Solaris. They all have process management functionality (which is what Solaris is introducing with N1 Containers in this release). What would possess me to use Solaris (which costs) instead of Linux or FreeBSD (which are free)?

It's like the mainframe folks tell you: If you know what you need it for, you'll be on the phone ordering it. Of course it's this attitude that got windows in the door for folks that didn't need minis, let alone mainframes, and forced the thin end of the wedge in. If you're word processing, you stick with Linux. If you're tracking audit compliance for 10,000 brokers in realtime from live, and you're doing it with an async I/O package, your programmer should probably be writing to an API and library that isn't labelled "WARNING: LARKS VOMIT", with half the functions documented as "TODO" and the source marked "hm, it seems to panic whenever you have x readers on y sockets, but if we fiddle here i think it could be ok with only a 5% chance of locking up completely". There's your robustness example, and I picked async I/O because it's precisely one of those areas where a lot of the free unixen fall down (it's actually one of NT's strengths too, but let's not talk about where NT falls down shall we?) Other areas might be partitioning (MOSIX really might not be your bag, depending on your app) or just plain features you want (doors, STREAMS, etc). There may be linux analogues to all these, but sometimes it's just Solaris's implementation that does the job best, or at least in the most scaleable fashion.

Re:Is Unix Unix?

[Anonymous Coward]

SUN support for hardware and software is COMPLETELY different. Try calling in a critical software problem and see what response you get.

Re:Is Unix Unix?

[multipartmixed]

> with a grocery bag full of spare parts (gbic cards, if you care).
> the problem was solved in a total time of one hour.

I love A5x00s, great machines, easy to diagnose (SUNWstade and/or front LCD), easy to fix.. I have had major shit fail on those without a second of downtime or byte of data loss.

As for your grocery bag full of spare parts -- GBICs are only about half the size of a pack of cigarettes, and a full-to-the-tits A5200 will only hold two of them.

You're probably thinking IBs (Interface Boards), which connect from the backplane to the GBICs. Two GBICs per IB, two IBs per SENA. They are hot swap, and the coolest thing is that when you swap one in, the array automatically flashes the firmware on the IB before bringing it online!

Oh, and as for GBICs, you should always have two spares on hand per array. The LED lasers in them are like light bulbs, when they blow it tends to be when you powercycle the units -- it's not a common occurence, but it's not rare, either. They only run about fifty bucks US a pop if you don't buy them directly from Sun. Try and get IBM GBICs if you can, some of the earlier Vixel GBICs were recalled (search the Sun FINs if you're buying Vixel to make sure you don't get a defective rev.)

Re:Is Unix Unix?

[Anonymous Coward]

I've been in both the Solaris and Linux environments as well, and I'm telling you now that the Achilles' heel of Linux is that it lacks the software/hardware integration and support that Sun (as well as IBM, HP, etc) can provide. Right now I'm being given the "Mom said, Dad said" routine between Dell and Redhat with a RAID issue. With Sun, they are mom, dad, your uncle, and your brother too. (and all that other recursive gene pool stuff)

The point: if you are balls to the walls mission critical (i.e. being down for an hour costs you 100k worth of business), Linux (and for that matter Microsoft) aren't even in the ballpark. You just take your lumps, pay for the support and get the appropriate support contract.

Re:Is Unix Unix?

[Frodo Looijaard]

strace

Re:Is Unix Unix?

[multipartmixed]

s/hold two of them/hold four of them/

Re:Is Unix Unix?

[Jim_Maryland]

I've always had good Sun support, pretty much tops compared to any others I've dealt with, although SGI runs a close second. On the PC side, Dell isn't too bad, but they do seem to move at the speed of the shipping company in delivering parts. Their phone support isn't quite as nice as Sun's, but the help is adequate in most cases (just need to get past the first level support staff).

The support problems you mention may be geographically related (not that that should be an excuse in the global workplace).

Re:Is Unix Unix?

[Anonymous Coward]

Well, no, but my ass deserves much better than that raging cunt.

Re:Is Unix Unix?

[Anonymous Coward]

i have worked in two shops in the last four years. one is a red hat shop. we use rhel es with paid support. the other was a full-meal-deal sparc/solaris shop.

Dude - no wonder you can't seem to hold on to a job. I'd fire you, too, if it you worked for me for two years and still couldn't use a "shift" key properly.

Re:Is Unix Unix?

[Veridium]

I can't, I don't work with Solaris anymore. But it's worth noting that in my 2 1/2 years working with Sun, we never had a critical software problem with their software. And as the guy from the Aus university made me realize, this was back in the 98-01 time span, so things might have completely changed since then. Also, I did ask if anyone had experience with x86 support, as that would seem more likely to expose someones experience with the software only side of the support(I'm saying this based on me only knowing of people who ran Solaris X86 on systems they built for themselves).

Re:Is Unix Unix?

[Anonymous Coward]

Well, dude, the wife's not gonna wanna get in bed with ya if you've got Klingons off Uranus...

Re:Is Unix Unix?

[forq]

Sounds like you had environmental problems. To lose 60% of your equipment, you probably had cooling or power problems. I have been working in the field for years, and Sun equipment is rock solid.

Only when we allow a particular area to overheat do we see increased failure rates in memory and CPUs.

We -do- lose power supplies and hard drives on a regular basis, but we're talking an installed base of ~1000 servers, and we lose 1 maybe 2 power supplies a month, and 3 or 4 hard drives a month.

Properly designed and built though, these do not cause any downtime at all. Mirror your drives, and properly route power from multiple sources to multiple supplies. N+1 is your friend.

And if you found Sun unhelpful, I would suggest you didn't feed them the right information. Install Explorer on all your machines, and use their pro-active support features that you are paying for.

Re:Is Unix Unix?

[jbplou]

Or you could have the Sun hardware which is the best on the market from a reliability stand point.

Re:Is Unix Unix?

[Anonymous Coward]

The main reason is hardware and warehouse consolidation. All systems if possible are moving to Dell (both workstation and server). By migrating to Dell and Linux we can keep our unix infrastructure while reducing the extra storage costs and spare part costs.

We also benefit from having a single hardware vendor point of contact (looks like software contact too as I am not even bothering calling RH again before calling Dell).

Before you ask, we had looked into migrating to Solaris X86 a few years ago. Recall what Sun did? Yanked the rug right out from everyone on Solaris X86. All our time (and cost) burnt at once. From that point on it was Dells with RH.

Re:Is Unix Unix?

[Anonymous Coward]

One key difference I have noted is the Real Time scheduling class in Solaris seems to work better for soft real time than those in the various Linux distros. That is unless you want something like MontaVista which provide true RT response, but limits the API. The latencies are getting better all the time in Linux, but at least in my experience, you sometimes get some huge (50 ms or so) for interrupt latencies. We never see these with the same code on Solaris.

Re:Is Unix Unix?

[deek]

• Your post basically sounds like exactly what someone who doesn't know what they're talking about would say.
  
  You don't write for ZDNet, do you?
  

Your post is lacking in anything that is informative and interesting. Your rebuttal consists only of base criticism unbacked by fact. Plus, you're an Anonymous Coward.

You aren't a Slashdot Poster, are you?

(Apologies to the many Slashdot posters who _are_ insightful in their posts.)

Re:Is Unix Unix?

[Anonymous Coward]

Try calling in a critical software problem and see what response you get

The response i get is "ftp the coredump to this site with the help ticket in the name of the file". A couple hours later i get a call back telling me what's up. Not always a solution that quick, and not that i expect one as the solution to a software problem should be WELL thought out.

Re:Is Unix Unix?

[deek]

• Linux was good good enough for Amazon.com back in 2001. They replaced SUN and HP with linux.

Unfortunately, the Amazon thing wasn't as simple as you make it out to be. Linux replaced specific servers for specific tasks that didn't need the enterprise-level features. They did NOT replace the core HP Unix servers, as they had the enterprise features needed for that part of the system.

Read this old article. Specifically, the most relevent part of the article is:

• HP helped Amazon migrate its customized software from the earlier servers to the Linux servers that dish up Web pages as well as to higher-end HP Unix servers for the heavy-duty systems nearer the heart of the operation, Balma said. "They're basically an all-HP shop."

So you see, they kept their "higher-end HP Unix servers". The reason why? Because Linux was not suitable for that job! With 2.6 though, we're talking a different story, but it's not field tested enough to go around and start replacing core servers with.

Re:Is Unix Unix?

[silence535]

Scuse me, but that card alone was worth something at the time the machine was brought. Namely roundabout 800 Bucks.
Buying top notch hardware today which you are not going to use today but five years later can only be called a waste of money.
Care if I gave you my bank account number to waste some money on? I swear I will buy some hardware and put it in the basement!
*SCNR*

-silence

Re:Is Unix Unix?

[rdr2]

Ya but then again do you remember the support run-around that sun gave about the gbic issues they had and also the 400 MHz Ultra Sparc II chip flaw? Basically they never admited to the problem that was widely know, instead they treated their customer base like they were idiots. Sure RedHat has its problems with support just like Sun, HP, IBM. But I would rather use the huge linux community on the web for my support, before I pick up the phone or submit a web ticket. I usually find an answer to my issue about 90% of the time.

Re:Is Unix Unix?

[Anonymous Coward]

Basically, for the enterprise, Linux wasn't really an option until 2.6.

I seem to recall hearing the same thing about 2.4 back in the 2.4.2 days.

As for why you'd want to use Solaris over Linux, nobody does NFS better than Sun. I'm not sure what the current status of NFS is under Linux, but I've heard some stories that don't look favourably on Linux and NFS.

There was the story in Linux Journal about ILM using a Solaris machine to serve NFS for a Linux render farm, but supposedly the NFS issues were worked out around 2.4.17 (give or take).

Re:Is Unix Unix?

[christophersaul]

Shame, considering what's happening now with Sun in terms of Solaris x86 and the lower cost Sparc kit!

Re:Is Unix Unix?

[Anonymous Coward]

The main reason that Linux is proliferating in the public/commercial sector (read not just geeks playing) is that it is a flavor that is running and supported on cheaper hardware - simply put - I am not forced into a hardware decision when it comes to running UNIX. For SMP - there is no better than Solaris.

Hardware?

[cujo_1111]

Sun has also added a new security tool with Solaris Privileges. This lets the root user create sub roots that can have permission, for example, to patch applications but not to touch hardware components.

How can an OS have control over a real, or non-virtual, object?

Re:Hardware?

[trentblase]

Haven't you ever opened your CD tray from your audio program?

Re:Hardware?

[Anonymous Coward]

Yeah, but I also can push the button on the front so that security sucks :)

Re:Hardware?

[Anonymous Coward]

More importantly, how could something be both new AND improved?!

Re:Hardware?

[Anonymous Coward]

That question has been asked many times before and is complete horseshit.

Something can be both new and improved at the same time.

new
To make new, to renew.

improve
To raise to a more desirable or more excellent quality or condition, make better.

Re:Hardware?

[trentblase]

Yeah, but I also can push the button on the front so that security sucks :)

but the OS can disable that button for many (if not all) drives.

Re:Hardware?

[Anonymous Coward]

Can it stop me from grabbing a screwdriver and dismantling the machine to get a hold of the fileserver HDDs?

Re:Hardware?

[Bastian]

$su
#chmod 600 /dev/sound
#exit
$mpg123 xxx.mp3

Re:Hardware?

[Trejkaz]

So _that's_ how the free drink holder works. Oohhhhhh...

Re:Hardware?

[Anonymous Coward]

Yeah, but if you were so correct, you wouldn't be posting anonymously. Oh wait... bugger.

Re:Hardware?

[lewp]

Yes, by delivering a mild electrical shock to your face. Difficulties in perfecting this process are why Longhorn's ship date keeps slipping.

Re:Hardware?

[LittleBigLui]

How can an OS have control over a real, or non-virtual, object?

HAL: I'm afraid i can't let you hot-swap that hard drive, Dave.

A simple question

[SoIosoft]

Can we expect Solaris 10 to be released to run on the x86 as well as Sparc hardware?

It's nice to have a real UNIX (Linux and BSD aren't really UNIX) that is relatively cheap will run on inexpensive x86 hardware.

Re:A simple question

[Brandybuck]

Linux and BSD aren't really UNIX

In all but name, BSD is every bit as much UNIX as Solaris. In fact, Solaris's precessor, SunOS, was directly derived from BSD. If you're hung up on names and trademarks, than BSD is not UNIX. But in every other sense it is.

Re:A simple question

[Anonymous Coward]

Solaris is based off SysV UNIX now, however. There are very noticeable differences between BSD and anything based off SysV, including Linux. And when I say that, I don't mean it's based off SysV code, but based off the specification.

And Linux isn't UNIX, either. If it were, it could possibly be bad for Linux in SCO's case against IBM.

I understand what you're saying, but you're not quite right.

Re:A simple question

[kscguru]

More important than just x86, it's going to come out on x86-64 (AMD). Anyone remember the two or three DUPED stories a month ago on the topic? *cough cough*.

It means rock-solid 64-bit UNIX on commodity x86 hardware. Very cool...

Re:A simple question

[Brandybuck]

Solaris is based off SysV UNIX now, however.

What difference does that make? There are several "official" UNIX systems that are BSD instead of SysV. AIX is one example.

Re:A simple question

[Anonymous Coward]

Yes, but AIX has also licensed UNIX code from SCO, although this license has since been revoked. Linux and BSD do not contain any of this UNIX code. Well, if you ask SCO, Linux might, but they have no credibility. *shrug* I know it really doesn't make a bit of difference one way or the other.

Re:A simple question

[Anonymous Coward]

Ask anyone associated with BSD and they will tell you they're not UNIX. BSD doesn't contain any of the UNIX code that something like SunOS (or even AIX which is based off BSD) does.

The same goes for Linux and why it's not actually UNIX.

And then there's also the trademark dispute.

Now go back in your hole. :)

Re:A simple question

[Anonymous Coward]

um... Since SunOS was derived from BSD Unix (not to be confused with netBSD) then they would have both been very similar at one point in time. That said, Solaris has had years and billions of r&d dollars poured into it. One might think that it would be a pretty solid OS.

Re:A simple question

[Anonymous Coward]

Only in the strictest sense of not being able to use the trademark.

Linux and BSD both do a pretty good job of being standards compliant despite not having been put through certification to use the trademark.

Some real (trademark using) Unixes are (or at least were) based on the BSD code base more than the SysV code base.

And if you look too much into standards...well, I'd say that in many cases, it's sad that POSIX chose the SysV rather than BSD ways of doing things (e.g. times is standard rather than getrusage; in practice all modern systems provide both interfaces, but SysV derivatives don't provide as accurate information as BSD derivatives).

Re:A simple question

[Anonymous Coward]

> What difference does that make?

The "Single UNIX Specification" is mostly modeled on System V. Modern BSDs like FreeBSD have spent some time adding System V APIs along side the traditional non-standard BSD APIs.

(IIRC, AIX is actually based on System III, with additional SVR4 code licenced from SCO).

However, that has nothing to do with the parent's point, which probably had more to do with things like init scripts.

Re:A simple question

[Anonymous Coward]

The Open Group, which does the UNIX certification testing, actually put Linux through the wringer and found a number of problems. And FreeBSD is probably even further off. The POSIX threading code in both OSes is brand new, for example.

So, Linux is not standard compliant enough to be branded as UNIX.

Obviously, TOG is hoping that someone will want to pony up to fix Linux and get it certified, as that's how they make their money.

Re:A simple question

[Znork]

"It means rock-solid 64-bit UNIX on commodity x86 hardware. Very cool..."

Not really. It means a rock-solid 64-bit unix that you should be running on SPARC. And by the way, it's gonna be discontinued next month. Or wait... no, we're gonna support it. Maybe. Or maybe not. You should be running Solaris on SPARC. But wait, you can run it on x86-64. Or SPARC. But we're going to discontinue Solaris on anything but SPARC next month. Or not. Well, hey, run x86 Solaris! No, it's not supported. Yes it is. Etc. Etc. Etc.

The fact is, until Sun can get their story straight for 6 consecutive months I wont ever consider running Solaris on anything but SPARC. As long as they cant commit for more than the attention span of a stoned gnat with split personality syndrome I have serious doubts about both the stability and the level of support one will recieve for non-SPARC platforms.

Re:A simple question

[ogre57]

Okay, tell me why SunOS is "real UNIX" but BSD isn't...

On the off chance this is a serious inquiry, SunOS is officially "branded", *BSD and Linux are not. Said brand is required to be a "real UNIX", costs $$$ to obtain. Vaguely recall reading that some Linux distro was going to try for this. Haven't heard of them in quite a while now (iirc it was with kernel 1.2!).

Re:A simple question

[Anonymous Coward]

Bill Gates, is that you fudding all over yourself again?

Re:A simple question

[MosesJones]

Can we expect Solaris 10 to be released to run on the x86 as well as Sparc hardware?

Quick answer: yes
Long answer: YES

NB The above only works in proportional fonts.

But to add more. Sun are supporting Solaris x86 as standard going forwards and they have a strong alliance with AMD. Probably the best way to thing of Solaris x86 is as Solaris AMD, sure it might work on the old stuff... but it will fly on the AMD stuff.

One interesting enterprise question about N1, Solaris 10 etc is this.... right now all of the software vendors charge per-CPU. With N1/Solaris 10 you can move to capacity based SLAs.

How will Oracle charge for the DB when you can't say how many CPUs its running on ?

Re:A simple question

[Anonymous Coward]

AIX license has not been revoked because Novell intervened and waived SCO's attempt to revoke it. SCO is only agent for Novell in SVRx licenses business.

Re:A simple question

[steeviant]

This article seems to show that Sun have made a very strong commitment to support Opteron based servers.

To quote; "The move does send a message, though, said Illuminata analyst Gordon Haff. "It certainly shows they're bloody serious about Opteron," he said, adding that it's likely Kealia's 59 employees are probably the largest concentration of Opteron specialists outside AMD."

I'm not going to argue about it being a good idea to wait until Sun have sorted out their operating system on the new platform though.

Re:A simple question

[jsavit]

You bet - Look at http://wwws.sun.com/software/solaris/10/

Re:A simple question

[__past__]

Yes, but having made very strong commitments hasn't stopped Sun before. Whatever their plans really are, people have a hard time trusting them after the back and forth regarding Solaris/x86 and their Linux strategy. It won't be easy for them to find early adopters in the enterprise space.

Re:A simple question

[Anonymous Coward]

How will Oracle charge for the DB when you can't say how many CPUs its running on ?

Same way IBM charges for DB2 on OS/400: by capacity unit.

LPARs aren't new.

Re:A simple question

[multipartmixed]

> UNIX systems that are BSD instead of SysV.
> AIX is one example.

Let me get this straight. You believe that AIX is based of BSD UNIX?

I'm not sure if you're aware of this, but SCO is suing IBM, claiming they copied SVR4 code from AIX into Linux.

I know it's not common knowledge, but you can read all about it at this nifty website I know.. I forget the name.. Its slogan is News for Nerds, Stuff That Matters... SlapDash, or something like that. Maybe you can find it with Google.

Re:A simple question

[Znork]

Acquiring 59 employees shows they're bloody serious? Mr Haff, the analyst, certainly has an interesting take on the concept of being bloody serious. 59 employees for a company of Suns size is probably less than the average weekly turnover.

Dont get me wrong, I like Sun. But in the x86 space they have _serious_ comittment issues, and they've been going back and forth in years.

If you want my 'analysis' I suspect that the Solaris Opteron idea is a strategic plan to try to derail the Itanium train somewhat more, and to one-up MS in the x86 game. That leaves them free to peddle Sparc in the high-end as the Opteron is no threat there for a long time, and it fits perfectly into their "_Solaris_. On. _Sparc_." overall strategic plan.

Which, of course, leaves any Solaris x86-64 customers high and dry as it's just yet another iteration of Suns "support-x86 but only when it suits our marketing and it's not really a serious product" line.

With that in mind, unless Sun can show a serious long range product plan and strategic shift into the x86 space for a longer period, I think I'll go with one of the several competitors who actually mean buisness when they're talking x86 and/or Linux.

Re:A simple question

[Brandybuck]

Every UNIX book distinguishes between BSD and SysV flavors. But no UNIX (in name or otherwise) is 100% BSD or SysV. That's because there's this thing called "POSIX" containing BOTH BSD and SysV APIs and facilties.

I'm not sure if you're aware of this, but SCO is suing IBM, claiming they copied SVR4 code from AIX into Linux.

Their claims changes from day to day. I would hardly trust a company that publicly displays a portion of 4.3 BSD source code and calls it their proprietary property, to be the final arbiters as to what is "UNIX". Last I checked, JFS and NUMA were not in SVR4 for IBM to "steal".

Re:A simple question

[evilviper]

On the off chance this is a serious inquiry

It was...

SunOS is officially "branded", *BSD and Linux are not.

Yes, I knew that.

However, the poster was saying it was implying that, somehow, Solaris was better than BSD or Linux by being "real UNIX"... A trademark difference most certainly does NOT fit that reasoning.

hmmm...

[Anonymous Coward]

I wonder is my old SS5 can still run it?

Re:hmmm...

[javiercero]

No, only 64bit kernels are provided now. So that means Ultra 2 and up type of machines are supported, Ultra 1 and the Sun4c/m/et al are now dropped.

Therefore Solaris 9 is the last stop for the sun4m machines.

Re:hmmm...

[g2racer]

Does this mean only an AMD 64-bit version for the x86 platform?

Re:hmmm...

[javiercero]

I dunno about the x86 version.

As far as Sparc goes, they are only supporting UltraSparc II and up machines. Seems that their sparc kernel will only be 64bits.

Some people were running newer kernels, with old drivers from the 2.6 days, these drivers were 32bit so i don't know how this will play out. (drivers for sparc/sbus stuff not intel stuff).

Sun tends to support their OS and HW for a loooong time, heck they are still supporting 2.6. So for most people 2.9 will be OK.

Re:hmmm...

[eatfrog]

The U1/U1E's were 64bit machines.
The problems is that they use Ultrasparc I cpu's.

Re:hmmm...

[sander]

You can run a 64 bit kernel on an Ultra 1. Its just not the default due to some silly CPU bugs that potentialy let somebody do a F00F style attack.

Re:hmmm...

[Probashi]

Ultra 1 is also 64-bit with correct e-prom and it is a sun4u not a sun4m.

sub roots

[gokulpod]

Sun has also added a new security tool with Solaris Privileges. This lets the root user create sub roots that can have permission, for example, to patch applications but not to touch hardware components.

This is a very interesting feature. Except for using sudo, does anyone know of any effort going on in linux to provide a similar feature ? Maybe Sun can port it to linux just to prove how OSS friendly they are ;)

Solaris vs. Linux

[bazik]

Solaris is great for the big Sun (Ultra)Sparc servers, but for the "smaller" machines with less than 32 CPU, Linux works so much better and faster. Not to mention the bigger choice of more current Software.

But then again, I might be a bit biased in my opinion :)

Re:Solaris vs. Linux

[Anonymous Coward]

What a load of BS. What is even sadder is that it was modded up so high.

Say there, you got Oracle and DB2 just a humming on Gentoo Linux on the 1-31 CPU sparc systems?

How about that EDA/CAD software used to build 100 million transistor computer chips? You know, the stuff that is massivly regression tested against Solaris since one tiny error can cost you a couple of million dollars?

How about commercial transaction processing middleware? Let me guess ... no?

What about peoplesoft?

SAP or any other Enterprise Resource Planning software?

Shop floor control?

Airline reservation systems?

I know: Web servers! Ya! 16 CPU web servers. And perl!

Re:Solaris vs. Linux

[g2racer]

Your argument is lost on me...

Oracle and DB2 (and let's not leave out Sybase and Informix to boot) both support Linux. Maybe not 30+ CPUs, but the software is definitely there, just need a kernel that supports massive SMP.

Don't know about commercial CAD/CAE packages, but my guess is if they're not there yet, they will be within a year or two.

Tuxedo and MQ are also available for Linux, arguably the industry leaders in TPM and middleware messaging. And in terms of app server technologies, all the vendors now have J2EE implementations that's as well supported on Linux as any other platforms.

SAP also is available on Linux. PeopleSoft too.

I believe that embedded (the kind used for shop floor control) is growing at an enormous rate (it's displacing real UN*Xs so fast as to render them irrelevant soon).

And unless you're an airline...

I used to be a UN*X bigot, but I've come to the conclusion (like IBM, Sun, and just about every other commercial software company) that Linux growth at the expense of proprietary UN*X is not necessarily a bad thing.

Re:Solaris vs. Linux

[kshcsuf]

You're missing out on some of the Solaris enterprise features if you use Linux on midrange Sun boxes. I know I pointed this out in a previous post, but I'll repeat it here for the sake of discussion. Solaris 9-current has system call performance this is equal or better to Linux 2.4.22+ in "all the right places". Not to mention, being able to use DR to add or remove system boards, CPUs, memory, disk and network interfaces on the fly in multi-user mode. I/O multipathing for fiberchannel and TCP/IP network multipathing--all carrier grade and "enterprise" ready. Running Linux in an enterprise setting on midrange Sun hardware seems like it would be a waste. You'd lose all of the above... granted, getting a working build of KDE on Sun hardware is significantly easier with a Linux variant. ;) ;)

Re:Solaris vs. Linux

[Anonymous Coward]

Oracle and DB2 (and let's not leave out Sybase and Informix to boot) both support Linux.

Tuxedo and MQ are also available for Linux..

SAP also is available on Linux. PeopleSoft too.

Actually everything you list is available for RedHat, maybe SuSe too. But none of it is supported on "Linux" and you'd get laughed off the line (Rightly so!) if you called any of those companies and told them you were running their software on Gentoy.

Re:Solaris vs. Linux

[Jonathan Platt]

Gatta call you on that one. Have you used Solaris?

The Ultra Sparks are a relatively slow processor compared to some of the others. Sun were even looking at a merger with Fujitsu a while ago, which would allowed Sun to drop the Ultra Spark and pick up the faster CPU, the SPARK64 (fujitsu) as their chip of choice.

The only reason Solaris (and Irix for that matter, which has the same problem with MIPS CPU's which are yet to break the GHz barrier) has really survived IMHO is because it is such a clean OS and is very efficient, so it can deal with slower chips.

I admit I havn't been using unix as much as a few of they other guys here, but that has been my experiance at least.

Re:Solaris vs. Linux

[Anonymous Coward]

Your argument is lost on me...

If you go back and read the parent and grandparent posts the topic was gentoo Linux on Sparc systems with 1-31 CPUs. As a practical matter there is virtually no commercial software on that combination on OS and hardware. Almost all of the growth in commercial Linux software is on Linux running on X86 with a teeny bit on Linux on PowerPC. As a practical matter the grandparent was an anti-Solaris troll by (someone claiming to be) a Gentoo developer.

Re:Solaris vs. Linux

[ReallyQuietGuy]

Spark

Erm. were you composing this on MS word or something with a spelling autocorrect that doesn't recognise SPARC?

Re:Solaris vs. Linux

[jbplou]

I would think that the smaller Sparc machines would perform just as well as x86 Linux machines as long as the applications they were running depended on memory more than processing power.

That rocks ....

[jobbleberry]

I loved the first movie, I hope they recast George Cloony, but I didn't like the other 8 ....

Uh wait this isn't about the movie Solaris is it?

Re:That rocks ....

[trentblase]

Yeah, but the movie was pronounced "So-lah-riss" and the OS is pronounced "So-lare-iss". That bugged me throughout the whole movie.

Re:That rocks ....

[cujo_1111]

How do you pronounce 'solar'? 'So-lah'? If so, then wouldn't 'Solaris' be pronounced 'So-lah-riss'?

If the OS is to be pronounced 'So-lare-iss', shouldn't it be spelt 'Solairis'?

Re:That rocks ....

[PacoTaco]

Sun passed up perhaps the best product placement opportunity ever by not getting their logo in that movie.

Re:That rocks ....

[tunah]

I loved the first movie, I hope they recast George Cloony, but I didn't like the other 8 ....

You mean the other ten? Oh, THAT movie.

Re:That rocks ....

[Anonymous Coward]

Our Sun rep says "So-lah-riss". Just FYI.

Re:That rocks ....

[Foddrick]

I always thought it was pronounced slow-lar-is.

Re:That rocks ....

[tr0llb4rt0]

Slow Loris

Apologies :-D

Re:That rocks ....

[trentblase]

If I was going by the pronounciation of "solar" I'd end up saying "so-luhr-iss". Anyways, check out the various videos at Sun's website. Here's one: http://www.sun.com/aboutsun/media/features/nc04q1. html McNealy says "So-lare-iss"... and he's the man... soo.... yeah.

Re:That rocks ....

[Mick Ohrberg]

That whole planet they're hovering over is running an ancient version of Solaris 1.x, and it's affecting their minds.

I wonder if Solaris 10 will run on my Ultra 1.

Re:That rocks ....

[mandolin]

I loved the first movie, I hope they recast George Cloony, but I didn't like the other 8 ....

Other 8? It's just the sequel, they're counting in binary.

Solaris 10????!!!!??!?!?

[_ph1ux_]

I just recently saw Solaris. Not even a sequel out yet... how could they be coming out with a 10th version?

It was good I guess, but good enough for 9 sequels in 3 years??

Re:Solaris 10????!!!!??!?!?

[evn]

They're numbering in binary.

Slow Solaris Upgrades

[theguywhosaid]

sure, some people are running solaris 8 still, by the cs dept here is running five .8

Re:Slow Solaris Upgrades

[theguywhosaid]

correct me, SunOS 5.8

Re:Slow Solaris Upgrades

[oob]

SunOS 5.8 = Solaris 2.8 = Solaris 8

Blame the marketing department.

Re:Slow Solaris Upgrades

[ogre57]

sure, some people are running solaris 8 still, by the cs dept here is running .. SunOS 5.8

Solaris 8 is SunOS 5.8, 9 is 5.9, 7 is 5.7, 2.6 is 5.6, etc. Guessing Solaris 10 will be SunOS 5.10. Part of why, pre-Solaris was 4.x so Solaris became 5.x, for eg version testing by scripts.

Other, have noticed that for whatever reason several companies deployed the even numbered Solaris versions, mostly skipped the odd ones. Meaning they were on 2.6, played with 7 a little, upgraded to 8 soon after it came out, have only played with 9. Seems they are treating it as if it were the Linux even/odd release/devel scheme.

Re:Slow Solaris Upgrades

[Anonymous Coward]

So in Sun's math, 2.8 > 7? I think I need to go lie down for a while...

Re:Slow Solaris Upgrades

[Anonymous Coward]

Well, 7 outright sucked ass, 8 seemed to correct a number of screwups, and as far as 9 goes, I dunno. My guess is there wasn't enough additional functionality to migrate. Anybody who has ever worked "change control" at a server farm knows you don't migrate for the fun of it.

We'll see if 10 is worth the pain.

Re:Slow Solaris Upgrades

[Anonymous Coward]

here's a handy table that lays it all out.

selinux

[Vic]

SE Linux is being included in upcoming releases of Fedora Core, and eventually Red Hat.
Link

"Solyaris" by Tartovsky

[mumblestheclown]

Wow. Solaris is so far off the radarscope of present computing these days that my first reaction to the headline was that there was going to be a followup to that clooney remake of the tartovsky film and that for some reason it would be cleverly numbred in binary (ie 10). Maybe this is funny (maybe not), but that was my honest first impression.

Solaris: all the craptitude of unix with all the benefits of vendor-lock in and high prices!

Re:"Solyaris" by Tartovsky

[Anonymous Coward]

Maybe this is funny (maybe not)

not.

Solaris: all the craptitude of unix with all the benefits of vendor-lock in and high prices!

Hey, at least it's not bastardized Unix. Unlike *cough*Linux*cough*.

Re:"Solyaris" by Tartovsky

[jay-be-em]

"Solaris is so far off the radarscope of present computing these days"
You obviously have nothing to do with present computing these days. At least outside of your basement. Oh, by the way, mom called, dinner is ready.

Re:"Solyaris" by Tartovsky

[Anonymous Coward]

I hope she's serving up the same thing she did for me last night.

This guy is right -- Solaris, and all the Sun hardware, is one generation behind x86 and PPC/POWER hardware with all their OSes (Linux, FreeBSD, AIX, even Windows if you discount the constant, disruptive security updates and look at the technology).

Re:"Solyaris" by Tartovsky

[Anonymous Coward]

That's nice but you've just confirmed the point. Outside your house, where people use IT for real business (and I don't mean running computers in your mates work with 30 staff) Sun's very much on the radar. Go out to a University or some other place with staff in the 1000's and you'll find plenty of Suns...

Re:"Solyaris" by Tartovsky

[Anonymous Coward]

if solaris on sun hardware is good enough for NASA, the CIA, the armed forces, countless medical and educational establishments, not to mention nearly *all* telcos and the world's stock exchanges, then they must all be "well off the radar" too.

I bow to your superior knowledge - and i'd like fries with that please.

Re:Solaris vs. Linux - mod parent up

[dtfinch]

I guess.

He's a Gentoo/Sparc developer, which makes him pretty well qualified to compare Linux to Solaris.

MOD PARENT DOWN, MOD GRANDPARENT UP

[Anonymous Coward]

Parent is a troll, grandparent is a valid question.

Re:MOD PARENT DOWN, MOD GRANDPARENT UP

[Peter Cooper]

Uh, THIS is Peter Cooper. If you're such a moron to think that an AC is me, then that's your problem, not mine. I don't hide behind AC. Clearly, you do. Know who you're talking to before acting like a dick. KTHX, bi.

Re:MOD PARENT DOWN, MOD GRANDPARENT UP

[wackybrit]

I think now you have been trolled, Sir. You are replying to a typical fuckwad AC troll. Best keep out of threads full of fruits like these. Damn, not taking my own advice.... :P

Arrgghh!

[jaguarxse]

Just when I finally decided to get certified in 9.....at least the upgrades aren't as prolific as with MS!

I've been running the beta for a while....

[Desmoden]

Has some cool features. Once apps (oracle etc) get "blessed" it will be nice to have a new core OS to go to since no one will support 5.9.

If for no other reason than getting away from a 101.5MB recommended patch cluster.

There are a lot of cool new commands for kernel info. There is also a performance increase depending on which cpus you are running.

Re:I've been running the beta for a while....

[Bitmanhome]

What's so scary about 102MB? That's like 3 Windows updates, or half a Debian update.

Re:I've been running the beta for a while....

[lewp]

Making fun of Windows and Debian in the same post should be every man's goal.

Re:I've been running the beta for a while....

[haggisman]

I think if you take the time to check the install log, you'll see that on average only about 20% of the patches are actually installed. The remainder are either already applied or don't have a base component installed. Thats the nice thing about doing it this way - you don't have to figure out exactly what patches you need. Ever read the patch documentation? :)

Why the hell is the parent modded troll?

[Anonymous Coward]

It's a valid question. And to answer it, yes, there is expected to be an x86 version of Solaris 10 at this time. Mod parent up.

Will they charge for x86 Solaris 10

[rueben]

And then change their minds a few months later, like Solaris 9?

Yes..

[Anonymous Coward]

Yes, Solaris will be released for x86 in addition to the Sparc release. And how is the parent a troll?

Solar is UNIX

[gnuman99]

Since Sun bought "rights" to UNIX from SCO some time ago, I think they can call it UNIX. Otherwise it would be Sunix and the President would need to change his/her last name to Sunis :P

Re:Solar is UNIX

[Znork]

Apart from the controversy over wether SCO can or can not sell any rights to their unix code at all, they most definitely cant sell the right to call something UNIX, as UNIX is a registered trademark which is not in SCO's possession.

So, no, the rights they bought from SCO do not convey the right to call it UNIX.

Re:Solar is UNIX

[Larry David]

You're speaking nonsense. You think that only people who bought rights from SCO can say their operating system is a UNIX?

You are confusing UNIX as trademarked noun, and UNIX as an adjective which can be applied to anything. The BSDs and Linux are both UNIX. Solaris is UNIX. SCO's UNIX is UNIX.

Re:Solar is UNIX

[James Youngman]

Since Sun bought "rights" to UNIX from SCO some time ago, I think they can call it UNIX. Otherwise it would be Sunix and the President would need to change his/her last name to Sunis :P

SCO has nothing whatsoever to do with whether you can call your OS "UNIX". That depends only on Unix certification. The Open Group also offers a list of cetrified products.

So is this version going to

[Anonymous Coward]

So is this version going to include the feature of it not being *fricking painful* to compile nearly anything not specifically targetted at Solaris?

No, I'm not trolling. Anyone who's worked with previous Solaris versions knows what I'm talking about. Anyone who's tried to compile GNOME as a non-root user on Solaris 9 is rolling on the floor crying from the memories right now. It seems like Solaris has everything just *barely* different enough that absolutely everything is a slightly different kind of complete pain to compile.

Yes I realize that at least part of this is that apps are targetted for Linux, so of course it isn't Sun's fault when shit doesn't compile. And yes, I'm exaggerating, the compilation problems only happen occationally, it's just that when they do happen it's really bad. But through the shit-colored glasses of memory, it seems like every time you try to compile some large free software package in solaris you uncover some new and painful oddity about the OS.

Re:So is this version going to

[Anonymous Coward]

I've never had more trouble compiling for Solaris than I've had compiling stuff for BSD. The basic fact is that unless software is tested on other platforms regularly, it gets very linux specific.

Re:So is this version going to

[Phibz]

I maintain packages for 300 or so programs for Solaris. I've compiled all of them using Sun's compiler, Forte from SunONE Studio 7. Although I agree that some programs are more difficult than others to compile under sSolaris, I've been able compile nearly anything I've attempted using forte 7. I used to use gcc but the speed improvements that forte adds make it very attractive.

I compiled GNOME and KDE and although I wouldn't say they were easy to compile I did get them working. And no I didn't compile any of it as the root user. I even was able to compile libavcodec something that supposedly runs on Solaris but is coded in a very very gcc specific way.

So I'm not really sure what difficulties you're refering to. So long as you have a sane build environment, gnu make, autoconf, automake, m4, a good compiler, gcc or forte, and know your compiler well you shouldn't have any problems.

Phibz

Re:So is this version going to

[plankers]

IBM has gone so far as to build a number of GCCisms into their native compilers, just to deal with this sort of thing on AIX. It's actually really nice -- most everything Open Source builds on AIX 5.2 now.

Re:So is this version going to

[Anonymous Coward]

Chances are some work is being done in this area, at least at the kernel/header level. They're doing a 64-bit opteron port, do you really think they've ported their entire development environment over to x86-64? Chances are they're using gcc, at least for the moment. Whether this encourages externally visible change remains to be seen.

Re:So is this version going to

[javiercero]

"Yes I realize that at least part of this is that apps are targetted for Linux, so of course it isn't Sun's fault when shit doesn't compile. "

That is the understatement of the year. Us BSD users have feel the pain of shitty code for years, as well as most other non Linux/GNU userland dependant Nixes out there.

I guess you have to take the good with the bad of OSS, the bad being the amount of shitty coders outthere that do not have a clue. And I have had my love/hate relationship with gcc for over a decade, sometimes I wonder why they still try to pretend to be a C-compiler :) and at least be honest enough and call it "new and improved C" or whatever but not C. And yes even with the strict ansi c flags I have had trouble with gcc. But I do not complain too much because I have gotten good things out of gcc too.

I just wish portability was real sometimes :(.

Re:So is this version going to

[aanantha]

Well, the thing that always pissed me of is the gettext incompatibility. Unless I disable internationalization most GNOME packages will fail in configure. GNU added extra functions to their own gettext with GNOME requires. I wasn't able to upgrade to Solaris 8 when I did this so I couldn't try out Sun's new gettext library. I heard that the GNOME programs will still demand the GNU one installed.

The other thing is I can never get the Jade stuff to work. The Openjade people didn't bother to mention that trying to build openjade as a shared C++ library causes core dumps on linkage. And even after getting that stuff installed it still doesn't work.

I'm amazed that you managed to zip past those without "any problems".

Re:So is this version going to

[Anonymous Coward]

Intel of course has done the same thing; ICC can now be used to compile a Linux kernel. A lot of the stuff in C99 came from GCC extensions originally, too. One might suspect that GCC is fast becoming the defacto standard in C compilers..

Re:So is this version going to

[aanantha]

I'm still trying to compile GNOME on Solaris *as* a root user. It's been a year now. Not done yet. Seriously. I wasted a lot of time trying to get that OpenJade crap working. I couldn't figure out why OpenJade would crash every time I ran it. It was only months later that I found out that a shared library version of OpenSP doesn't work on Solaris. The Openjade people didn't think it important enough to mention. But even after getting that stuff built and figuring out how to install DTDs by looking at my Linux system, it still wouldn't work. So then there was no choice but to disable all gnome doc building.

And then there's that goddamn gettext/libintl stuff. I don't have the option of overriding the Sun version of libintl across the board by placing the GNU libintl in /usr/local. Our Solaris software that uses Sun's libintl will break. You think the GNU people could have at least made it possible to use a different library name (libgintl maybe) for their incompatible version of libintl and have their configure scripts search for that.

Now all my stuff is out of date so I have to start over from the beginning again. Building GNOME has just become a great way to procrastinate. I wonder if I'll ever get to KDE.

Sun tries to solve this problem by giving you a prepackaged version of GNOME. But 2 problems with their approach: 1) it's never up to date, and 2) it's only good if you have Sun's C compiler. I don't know how, but somehow Sun's GTK+ C library is incompatible with GCC's. Same with their Zlib. So I can't use Sun's GNOME distribution to write software. I though that kind of stuff was only supposed to happen with C++. I'd much rather like Sun to promote a /usr/ports like thing for Solaris. Whatever they did to get GNOME to compile they need to stick those patches in a /usr/ports. The community would help out. Then eventually we can have full "distributions" that we can install onto a Solaris system. Life would be a lot easier if there was something like "fink" in place for Solaris.

And while they're at it, Sun needs to stop ignoring GCC. Make the C and C++ ABI compatible. There are a lot of Solaris software houses that only use GCC because they need to build their code on other unices. They're not going to be able to use Sun's GNOME environment if it's not GCC compatible.

Re:So is this version going to

[sander]

Get more losers^H^H^H^H^Hdevelopers to have a clue about portability? And no, it is a different thing from "runs on both suse and redhat on x86".

Re:So is this version going to

[tfb]

Well, we run 7 and 8 in production and 9 in development, and I have just a huge mass of free software. which I've built. I admit it's been a while since I compiled anything really huge (bigger than gcc say, and even there I think we're still on 3.3), but I don't often have problems. I have built gnome in the past and I don't remember it being a saga. I'd not bother building it now since it ships with 9, but I kind of assume it builds OK since it does ship with 9...

Maybe the issue is not using gcc? A substantial proportion of `C' software is actually *gcc* software, so you really need it. In just the same way a large number of linux `shell scripts' are actually `bash scripts'.

Re:So is this version going to

[dglo]

So is this version going to include the feature of it not being *fricking painful* to compile nearly anything not specifically targetted at Solaris?

I'm old enough to remember back when Sun was the top dog and all non-SunOS users complained about Unix software being written specifically for Suns.

And before that, the problem was people writing Unix software specifically for the VAX.

This says more about the quality of the people writing OSS than it does about the quality of Solaris. Actually quality is the wrong word, because it's likely that many of these programmers only know or have access to Linux.

Making an effort to port your software to non-Linux dialects of Unix is a really good thing, because other OSes will expose bugs in your software which would not otherwise turn up until, for example, a bug in the current glibc is fixed.

Re:So is this version going to

[Mr. Piddle]

Have you tried compiling the latest release of GNUCash? If you are a hard-core compiling masochist, as you claim to be, I highly recommend it.

Re:So is this version going to

[Jon_E]

agreed. The market and what people code towards only changes based on what hardware most people have access to. That said, affordable solaris systems for joe average developer working from his basement are typically only found on ebay .. it'll be nice if Sun can get S10 x86 (or Sparc) bundled with lower cost PC sales - for now it seems pretty limited to their volume servers like the V20z which are still a little pricey and not effectively aimed at the home/small business market. The Sun blade workstations are pretty nice, but the market tends to frown on the price:clock-speed ratio - and component-wise they've tended to struggle and hang onto their flailing graphics cards for too long.

For the developer, I've found the typical move to Solaris goes something like .. "this isn't what i'm used to, so it's stupid" .. to eventually (if they last) .. "oh, that's why they're doing it this way - this makes more sense" - often the reverse of what happens as you dive deeper into Microsoft, OS X, or Linux.

IMO the packaging has always been much cleaner and simpler than anything else out there, and the newer ideas tried and tested with lessons learned from huge enterprise environments tend to eventually make their way into Solaris first. For example, take a look at pre-emptive kernels, kernel memory allocators, threading models, networking optimizations .. and the list goes on. "Quality" comes from experience, and that experience tends to shift from age to age.

Re:So is this version going to

[Wolf Eyelash]

People need to be aware that Sun tries very, very hard to keep their ABI's stable. Certainly they are going to be careful with the libc ABI. Note, to help folks out Sun provides a Freeware CD with most of the tools (gcc, gnu libc, etc...) needed to build freeware source. So you really have the best of both worlds (a stable ABI and a evolving ABI).

Re:So is this version going to

[Probashi]

I was able to build Openjade on Solaris 8. But, I have to agree that there were a lot of hardles along the way. Makefile that the configure creates is not correct. Also, the installer does not copy all the dsssl dtds properly.

Re:So is this version going to

[aanantha]

But the Freeware CD is not going to be enough anymore. Once GNOME is integrated into Solaris, Sun won't be able to keep the GNU stuff separate. They need to address the incompatibility. And they need to make it easy for GNOME developers to test software on Solaris. They won't be able to afford Sun's compiler, so Sun should ensure that GCC remains compatible. As for the ABI, the C ABI really was supposed to be stable. When basic things like Zlib and GTK+ libraries are incompatible, you either have a choice of using Sun's GNOME distribution with Sun's compiler, or to throw it all out and rebuild everything with GCC. Once GNOME is integrated, you probably can't just throw it all out.

never do anything with the odd release of solaris.

[Desmoden]

think of it like odd linux kernels.

5.6 stable we all used it.
5.7 we played with, tested
5.8 we all upgraded to, used, liked.
5.9 play with and test
5.10 upgrade and enjoy.

most oracle products for example will never be certified on 5.9. It's too much work to requalify and upgrade to a new solaris version. So the odds introduce new features and work out the bugs, and the evens is what we use.

Re:Solaris vs. Linux - mod parent up

[javiercero]

Not really, actually I have had the opposite experience with Solaris running much better on desktop sparc machines than linux.

Also most of the software out there that can be compiled in linux can also be ported over Solaris with minimal grief.

And I do not particularly feel like spending 2 days compiling in order to have a stable machine. A solaris install with the extra software CD provides most of the functionality than a linux install. But if you like linux by all means go and use it. Saying that linux is somehow better or makes more sense than solaris just because is just plain dumb.

2.2 in Debian stable?

[Anonymous Coward]

So it's been quite a while since you checked, huh?

Or when you say "checked" do you mean "heard from a troll on slashdot"?

Re:2.2 in Debian stable?

[phrasebook]

But it does have 2.2 by default doesn't it? If I press Enter at the boot prompt of my woody CD, I get the 2.2 idepci kernel. So I would say that is the default.

But you can type 'bf24' to get 2.4 of course.

Re:2.2 in Debian stable?

[randomblast]

I tried Debian Woody on one of my spare machines, and i thought it was seriously out of date, it may have the option of a 2.4.18 kernel (the same as RedHat Shrike), but it's got KDE 2.2, which sucks.
I tried to go online and use apt-get to update it, but it didn't fetch any of the packages i wanted, apparently they are for the testing release only.
Ah well, wiped Debian, got my SuSE 9 with KDE 3.2, and it's compiling the 2.6.2 Kernel now.

there's an old saying...

[SuperBanana]

There's an old legal saying- never ask a question you don't already know the answer to.

The company is so confident about Solaris' speed that officials repeatedly offered to challenge Linux on benchmarks in the coming months.

Now, usually companies don't make such bets unless they're well hedged. So, perhaps running some benchmarks against the preview versions of 10(the article mentions most of 10 is available already to update subscribers) might be a nice idea, to see what's got Sun so cocky, instead of just saying "oh. Solaris is crap"(which is at least partially wrong anyway).

Re:there's an old saying...

[Anonymous Coward]

the article mentions most of 10 is available already to update subscribers

Actually, Solaris 10 is available to everybody through the Solaris Express program. And it's free (for non-commercial use).

Re:there's an old saying...

[jcupitt65]

I found this other article linked on the reg, talking about the new TCP/IP stack:

http://www.theregister.co.uk/content/61/33440.html

The relevant quote is (near the bottom):

We worked hard on efficiency, and we now measure, at a given network workload on identical x86 hardware, we use 30 percent less CPU than Linux.

So I guess that should give them a fair performance advantage under very heavy static loads. Although he doesn't say which linux.

Re:there's an old saying...

[jrockway]

Umm, microsoft claims that XP/2003 is more secure than Linux. In the wake of the last eight virus attacks and source code leak, I think M$ is wrong. Maybe I'm wrong, though? :)

Re:there's an old saying...

[kshcsuf]

There's a reason why they are piping up the rhetoric... very recent recommended patchkits include a kernel patch (dated around Christmas eve) that have very specific "fixes" which include massive performance improvements in over a dozen system calls. Each fix/report states that those system calls were slower when tested against Linux 2.4/2.6. The fixes have been back-ported to Solaris 9 and are included in Solaris 10. Hope this helps.

Re:there's an old saying...

[Anonymous Coward]

Correct me if I'm wrong, but don't most linuxes use the same TCP/IP stack code? If I'm right, and they aren't lying, it's safe to say that that claim is true for most linuxes.

Re:there's an old saying...

[mczak]

I think the OP was refering to the linux kernel version. In fact, given the date of the register article (october 2003) it's probably safe to assume the comparison is against some 2.4.2x kernel. AFAIK the 2.6 kernel though has a quite improved tcp/ip stack, though I'm not sure if it's actually faster or just a cleaner implementation...
So this comparison is meaningless, you might argue that 2.6 isn't quite enterprise ready, but neither is Solaris 10...

Why all the bad comments?

[KidSock]

Excuse me but this actually sounds pretty good. This "containers" thing permits running hundreds of virtual machines on one host (and not a moment too late as that idea is becoming a very popular -- I have a VPS runing UML and it's very snappy). The DTrace utility sounds nice although I probably shouldn't say that considering I've never tried it. And they're going to run Opteron and claim that they can beat the Linux benchmarks. I don't know about you but I wouldn't mind having an Opteron box running Solaris 10.

[disclaimer: I have 50 shares of SUNW]

Re:Why all the bad comments?

[Anonymous Coward]

50 shares of SUNW == $100 (US). Yawn.

Re:Why all the bad comments?

[Ziviyr]

If Solaris beats Linux, how long after the benchmarks do you think the kernel devs will take to reverse that? :-)

Re:You're late to this scene

[Bastian]

Solaris came out in 1972. There's been plenty of time for 9 sequels in 30-odd years. Actually, the crappy Solaris that just made the theatres in the USA is Solaris 9, but was released in the USA as Solaris 1. (It's kind of a Final Fantasy 3/6 thing) Solaris 11 is already in the process of being shot, just like Darl.

Re:Solaris vs. Linux - mod parent up

[dtfinch]

I just thought he was unfairly modded down to -1, Troll. His experience developing with both Linux and Solaris on a sparc should make him more qualified than the average troll.

Re:never do anything with the odd release of solar

[dtfinch]

Just like all the odd numbered Star Trek movies. Ask any trekkie which of them were the best, they'll often say 2,4,6,8,10 in whatever order.

Dtrace?

[Anonymous Coward]

The comments about DTrace are clearly ludicrous:

DTrace sends the probes through a server looking for hardware errors and anything that might be slowing application performance.

DTrace is a sweet tool for anyone who's had the chance to run Solaris Express, but a much better description can be found at the source.

Re:Dtrace?

[Anonymous Coward]

DTrace sends the probes through a server looking for hardware errors and anything that might be slowing application performance.

This is actually pretty sweet. You can learn more about it here

Re:Not just Solaris

[Bastian]

I've run into the same problem on Mac OS X. Usually you're fine, but when you aren't, you're in a world of pain. I think this is a lot of why fink just creates a separate file tree rather than trying to merge itself in with the main one.

I've actually been considering dual-booting GNU/Linux on my mac for just this reason.

Re:Solaris vs. Linux - mod parent up

[Anonymous Coward]

What do you have against trolls you fucking asshole. We're people too, Fuck I'm going to waste you.

timezones & reboot

[Anonymous Coward]

with this release will i be able to change the timezone without requiring a reboot?

Re:timezones & reboot

[twoslice]

with this release will i be able to change the timezone without requiring a reboot?

Yes, I do this all the time when I fly from Chicago to LA with my laptop.

Re:timezones & reboot

[Anonymous Coward]

with this release will i be able to change the timezone without requiring a reboot?

So "Funny" == "Stupid" in modworld?

Re:timezones & reboot

[tfb]

Only if you have a functioning brain. So probably not.

PARENT IS A TROLL! MOD DOWN!

[Anonymous Coward]

Re:my experience with slash-dot

[Anonymous Coward]

...but, but! office _IS_ E-Vil. Anyhow, girls are lame. I'd rather have a frog that talks.

Re:Solaris vs. Linux - mod parent up

[Anonymous Coward]

I had several sparc boxes of varying degrees, and they all needed OSes. Solaris installer would take 10 minutes to boot up CDE, just to tell me that it can't find a harddrive. Booting up RedHat 6.2 for sparc resulted in a quick boot, and it found the hard drive too.

Come on. Sun's OS doesn't recognize a Sun HD on a Sun computer, when Linux does. ......

Re:Solaris vs. Linux - mod parent up

[ajagci]

Not really, actually I have had the opposite experience with Solaris running much better on desktop sparc machines than linux.

Well, SPARC hardware isn't exactly mainstream software for running Linux or GNU utilities anymore.

Also most of the software out there that can be compiled in linux can also be ported over Solaris with minimal grief.

I have lived with Solaris and Linux side-by-side for a few years. It's wrong to say that "most" software can be ported over [to] Solaris with minimal grief; the fact that it requires any porting at all is already a lot of grief. Even for software using autoconf, compiling on Solaris fails frequently.

And I do not particularly feel like spending 2 days compiling in order to have a stable machine.

So, why do you tell people that it's a good idea to run Solaris and then spend days recompiling all their favorite GNU or Linux software? Because that's what it amounts to: every Solaris machine I have ever used, I had to spend days installing GNU software on it because it came with so little, and what it came with was usually far inferior to the OSS equivalents.

A solaris install with the extra software CD provides most of the functionality than a linux install. But if you like linux by all means go and use it. Saying that linux is somehow better or makes more sense than solaris just because is just plain dumb.

Spending money on SPARC hardware seems "plain dumb" to me, given its price/performance ratio, in particular if you are going to run Linux anyway.

Re: stronghold

[phr1]

Stronghold basically doesn't exist any more. It's just regular Apache with mod_ssl now. There stopped being any reason to maintain the old Stronghold module. And mod_ssl is included with Apache 2.0 by default. So yeah, RHEL comes with an SSL web server; whether they bother still labelling it "Stronghold" is not terribly relevant.

Linux Business Week says Solaris 10 has..

[Anonymous Coward]

...600 new features

Re:Linux Business Week says Solaris 10 has..

[Anonymous Coward]

Sure it shouldn't say 690 new features?

Re:Parent is troll

[Prof.Phreak]

But but but... GNU's Not UNIX :-)

Solaris, on the other hand, doesn't deny being UNIX :-)

j/k

Re:Solaris vs. Linux - mod parent up

[javiercero]

Well, when you can get a 100+ processor single image x86 machine let me know. Solaris/Sun still makes sense for some scenarios. If you want to run linux I agree that a sparc box is not the way to go, if you have a legacy box maybe. But with the 2.9 release solaris has beem quite responsive for ultra based desktops.

If you want to run your linux desktop, probably you may be better off with an x86 machine.

Also believe it or not, GNU is not the "defacto" for everyone using UNIX you know :). Whenever I need a gnu pakage I can just get to the solarisfreeware site, or use the pkg-get script and get it, install it, and be ready to run it within seconds. That sounds pretty simple to me. For the most part I like to keep some of the machines I use as GNU free as possible due to the dependency and bloat hell that it sometimes generates, I have to use commercial tools for part of my research.

Ironically, though...

[Artifex]

Considering Debian stable, last I checked, still has the 2.2 kernel as default, I'd say you have at least a ten year wait.

Ironically, though, if Debian is the new Slowlaris , that means it has it right now.

SELinux gives you this

[Oestergaard]

Solaris 10 contains the Trusted Solaris security features (labeled security, mandatory access controls (MAC)) which is what allows such flexible administration without the almighty root user.

I haven't run the prerelease of solaris 10 myself yet - but from what I've read, they have really taken the trusted solaris features and put them in solaris 10 - this is not just the RBAC features from solaris 9 (which would actually allow the described sub-root concepts, but not all the other goodies that come with real MAC).

This is what SELinux brings to Linux. You can run Debian stable with SELinux if you really want to. Otherwise, look for RH AS 3.0, or get to work on testing SELinux in debian unstable so that we can all get this functionality in the next debian stable.

Google around for selinux on debian and you should be able to find out how to do this.

Re:never do anything with the odd release of solar

[kshcsuf]

I disagree with the parent poster. 5.7 was the first OS generation with 64-bit support... including kernel modules that were compiled for 64-bit use. This yielded a significant performance boost for heavy I/O-based applications, as device drivers could transfer larger quantities of data with a single transaction (if using PIO).

Also. Solaris 9 incorporated priority paging and the new single LWP/kernel thread model. The new model makes large-threaded applications scale incredibly well.

All 35 of my servers are running Solaris 9 and they run very well.

Mod parent up (with SELinux link)

[AmVidia HQ]

http://www.coker.com.au/selinux/

Grandparent is technically right

[Larry David]

GNU is referring to the trademark 'UNIX'.

That said, the GNU initalism is a little out of date, since it was created to represent GNU as a final operating system.. which still hasn't happened (although Hurd is coming along nicely).

By 'GNU's Not UNIX' they mean.. GNU isn't 'UNIX' the evil monolithic system which was trademarked by AT&T. They didn't mean that's it's not UNIX-like.. or, as we say nowadays, just 'UNIX'... as in 'Linux is a form of UNIX', etc.

So no, Solaris is no more 'a UNIX' than Linux is 'a UNIX' or FreeBSD is 'a UNIX'.. They are all UNIX, but they are not all UNIX (TM).

advocatus diaboli

[Imperator]

Isn't this the sort of thing that we'd be up in arms about if it were MS compiler quirks other compilers were emulating? The GCC compilers should make strict conformance to modern standards the default, and make you turn on the extensions manually.

Why, for example, does GNU C++ include binary <? and >? operators for min and max? I could see the attraction in C, where preprocessor macros and their issues with side effects are a pain in the ass. But in C++, inline templated functions can do it just as well and are much more portable. This is the sort of irresponsible "extend the language by default" approach that the GCC compilers are full of.

Don't get me wrong; I love the GCC suite and for all the supposed performance issues, I wouldn't trade GCC for any other toolchain I've ever used. But free software should set an example by encouraging portable code.

</rant>

Re:advocatus diaboli

[Mr. Piddle]

But free software should set an example by encouraging portable code.

My theory is that RMS is really Bill Gates in disguise.

Re:advocatus diaboli

[Empty Threats]

Well, that's the thing: GCC runs on every platform I've ever had, from MINIX on a 16 bit processor to Solaris on a 64 bit processor.

MSVC (sometimes) works on (specific versions of) Windows 2000 (when it feels like it) (and the .NET framework is installed) (and .NET doesn't hose the system).

Re:advocatus diaboli

[plankers]

Yes -- actually there are a lot of people that are dissatisfied with this. You just don't hear the outcry because it has been transmuted to a collective "argggh" when people try building software that should be portable and it won't build. The idea that GCC is preventing the growth of open source is something just not propagated, because it's heresy. A lot of what GCC 3.x has done to adhere to the language has been wonderful, but that also means they did not maintain compatibility with GCC 2.95 and before. That's okay -- in my position I'll take the portable code that results from the move to 3.x, especially 3.2.

I love GCC, but it's just not the compiler for these other platforms. If you built all your software with it -- fine. But I find I need the native compilers (Forte or VisualAge) to be installed alongside it (for vendors and such), and what ends up happening eventually is a linker nightmare. Especially if you need to use libraries that were built with the other compiler. For the price of the native compilers, at least for me and my organization, it's not worth messing around with GCC on these systems.

Because nitpicking is fun.

[Anonymous Coward]

SPARC hardware isn't exactly mainstream software.

You don't say!

Re:ibm,hp,compaq *don't* make high end x86 servers

[Anonymous Coward]

You make it sound like ibm,hp,compaq did not make high end x86 servers. Himilaya non stop servers come to mind. Heck even proliants are nice.

They don't.

Not when compared to this beast. Each of those 72 UltraSPARC IV CPU chips is actually two 64-bit CPUs on die.

144 CPUs, 1/2 a terabyte of RAM, near linear scalability.

In fact, the Intel architecture pretty much can't scale well beyond 4-6 CPUs because of problems like cache coherency - i.e., when CPU a modifies a memory value any CPU that has the memory value in cache needs to know that cached value is now invalid.

SunOS to Solaris: BSD to SVR4

[Anonymous Coward]

That is in fact exactly what happened. SunOS was BSD-based. With Solaris, Sun switched from BSD-based to SVR4 based.

And on a multi-CPU box running Solaris 2.5, don't ever try the BSD way of allocating multiple pseudoterminals - use the SVR4 way of a master device (/dev/ptmx, IIRC) lest you panic your box...

all my replies in one go

[chegosaurus]

Wow some of the comments on here are uninformed. Especially those modded informative or interesting.

"Hey, Solaris sucks! Linux is way better and it's free as in speech!" +5 interesting

"Hey Linux does everything Solaris does and it's free as in beer" +5 informative

"Hey there was a film called Solaris! OMG LOL!!!" +5 funny

Do you *never* get bored of pointing out that x86 chips have higher clock speeds than SPARCs?

Don't you think we *know* by now that Linux is free?

If you know how to handle Solaris, you will know that: it has some features that linux does not. It's no harder to build software for than linux. Trusted Solaris privileges are not the same as sudo. dtrace is not the same as cat /proc/whatever. Solaris is not so slow it's unusable on 32 CPUs. Version 5.8 is not four versions older than 9. There are smarter ways of patching than downloading the recommended cluster every day. But hey, post uninformed crap and up your karma. That's what matters.

If you don't know what you're talking about, shut up and leave the discussion to people with some interest and background in the subject. And stop complaining that "no one uses Solaris, so who cares there's a new major release", when you've probably been up all night bitching on IRC that the mods here rejected the 2.6.3-rc3 release story you submitted.

Re:all my replies in one go

[sander]

Come on, this is slashdot, that unfortunately tends to serve as a site for losers who don't even run Linux themselves to pretend to be c00l l33t l1nux h4xors and pour shitty comments on everything that is in their small minds not cool.

Re:all my replies in one go

[Anonymous Coward]

Solaris is not so slow it's unusable on 32 CPUs ... If you don't know what you're talking about, shut up and leave the discussion to people with some interest and background in the subject.

Get upset about offtopic Linux posts, fair enough. But just lay off the attempted cheap shots at Linux. It just makes you look like an imbicile.

32 CPUs, eh? Well that must be a very funny scalability curve that Linux has, because it seems to come good again at around 512 CPUs. Jackass.

Re:all my replies in one go

[sander]

Maybe you should find out what 'STREAMS' is and what it measures? As a hint, basicly no interactions with the OS happen in the benchmark run - its all about memory bandwidth, measured in a NUMA scalable way.

You would have the same Streams score even if the kernel running on the machine used a monolithic giant lock - it does not say anything about OS - as opposed to memory hardware - scalability at all.

Re:all my replies in one go

[chegosaurus]

> But just lay off the attempted cheap shots at Linux

I never mentioned linux. I said "Solaris is not so slow it's unusable on less than 32 CPUs". (My bad here - I forgot I was putting stuff on an HTML message board and just typed a "less than", which disappeared, then missed it in the preview. D'oh!)

That, for the hard of thinking, means that I believe Solaris is not slow on small machines. It in no way implied that I think linux does not scale. (I don't know enough about linux on big machines to comment.)

> Jackass.

I'll lay off the cheap shots at linux if you lay off the cheap shots at people whose posts you can't understand. Deal?

Some 'scoop'...

[DrMindWarp]

...when you can look at Sun's website and read all about Solaris 10.

You can even download the beta.

pkgsrc may be what you're searching for

[cquark]

Only a few years ago, SPARC/Solaris was the most standard platform for open source software and IA32/Linux was the nonstandard, difficult one to build on. It's amazing how fast times have changed.

As for libraries compiled with a different C compiler than you're using to link with, that's a common problem between gcc and vendor UNIX C compilers. However, the vendor C compiler suites shouldn't be disregarded as they offer many advantages over gcc (take a look at some of the Solaris bugs in gcc and gdb.)

However, if you want something like /usr/ports on Solaris, check out pkgsrc. It's NetBSD's ports collection, and it has been ported to Solaris 8 and 9.

Re:pkgsrc may be what you're searching for

[aanantha]

I agree that the Sun C/C++ compiler is very good. But gcc is here to stay. If Sun is serious about adopting GNOME as their user interface, then they need to accept everything that comes with it. GNOME software is community developed, and the developers will not test the software on Solaris with Sun's compiler. They will only use gcc because it is free software. It is not tractable for Sun to do all the work verifying that each GNOME package compiles and operates with Sun's compiler. And it is not acceptable to assume that every Solaris developer will use Sun's compiler.

Sun's needs to do what other UNIX vendors have done. Actively provide fixes and performance improvements for GCC and do what's necessary to ensure binary compatibility. I'm not saying that Sun needs to break their own released ABIs, they could instead work to change GCC so that it becomes compatible with theirs. GCC users are already used to evolving ABIs. It would be worth it for Solaris GCC users.

Re:pkgsrc may be what you're searching for

[aanantha]

Thanks, pkgsrc looks really cool!

Re:never do anything with the odd release of solar

[sander]

You can use the single LWP/thread model in Solaris 8 too - just switch to the alternative libpthread.

Re:Not just Solaris

[Anonymous Coward]

Having just recently purchased an iBook and promptly wiping the drive to install linux, I can tell you it's just as much a hassle running PPC linux as trying to compile on Sparc Solaris or OSX.. AFAIK the big issue is a lotta people don't just write their code for linux, but for x86 linux, possibly due to endian issues. Forget what it was I was trying to compile, but at least a fourth of the programs I tried to compile on gentoo failed (just for reference gentoo includes a VERY small subset in PPC, so most of the programs I was compiling were x86 arch only)

-- vranash

Solaris?

[Omega1045]

Is this a new Sun distro of Linux?

Re:never do anything with the odd release of solar

[maitas]

http://sdc.sun.com/solaris_list/s9supported_prod _a lphaO.html

List of ORACLE releases supported on Solaris 9....

The main answer is three simple letters.

[devphil]

And they are: R. M. S.

This is the sort of irresponsible "extend the language by default" approach that the GCC compilers are full of.

So, you'll be submitting a patch, then. What's that? You say you haven't tried it? Ah.

We've tried getting rid of some of the extensions that were not well thought out. He's just dead-set on keeping them.

Also -- unfortunately -- many of them are actually being used. Pulling the rug out from under your users is not a keen move.

It's a slow battle. Many were removed for 3.4. More are being removed for 3.5.

But free software should set an example by encouraging portable code.

Heh. The FSF view on this is that GCC is far more portable than any of your code is likely to be, so just install GCC instead and make use of its features. Not a position that I agree with necessarily, but it is the same argument that I make when installing bash on all the non-GNU systems I use. :-)

Re:The main answer is three simple letters.

[Imperator]

We've tried getting rid of some of the extensions that were not well thought out. He's [RMS] just dead-set on keeping them.

But how much code does RMS contribute to GCC? Shouldn't GCC, like other OSS projects, ultimately be in the hands of its developers?

Also -- unfortunately -- many of them are actually being used. Pulling the rug out from under your users is not a keen move.

I fully understand that. That's why I suggest leaving them in but requiring flags to activate them. Then you can give errors like obsolete extension >? requires -fminmax-operators. If users don't want to change their code, they can add some flags.

Re:The main answer is three simple letters.

[devphil]

But how much code does RMS contribute to GCC?

After version 1.something, none.

Shouldn't GCC, like other OSS projects, ultimately be in the hands of its developers?

You'd think that. So would I. But GCC is not like most other OSS projects. Proprietary compiler writers would love to be able to take bits and pieces of GCC and get away with it. A huge step in preventing that is the fact that copyright is held by the FSF, not by the developers. (So if you steal the code that I've added to GCC, it's not me having to pursue a GPL violation, it's the FSF. Makes an even bigger difference when the parties involved are oversees.)

So RMS still has copyright (well, the FSF does, but no difference there) and therefore, essentially, the final say over major changes to GCC. (It is, after all, the FSF's flagship "product" these days, so he won't give that up.) Veto power, basically. He can be reasoned with... eventually... slowly... and the GCC Steering Committee is good at doing so.

Shoot, for a long time he wasn't going to allow the gcj people to bring their code into the master GCC tree, because he thought the portability of Java bytecode could lead to GPL violations. He's a bit out of touch, technically speaking, but he still holds the big stick.

The answer, in my (correct) opinion *grin*, is another EGCS-style fork. But as long as he is willing to make changes, even small and slow ones, the majority of the developers are loathe to take that step, and understandably so.

GNU/Linux as a High-End competitor...

[borgheron]

So long as Sun continues to ignore GNU/Linux as a serious high-end competitor to Solaris, GNU/Linux has an advantage.

Sun seems to be of the opinion, "If we ignore this GNU/Linux thing it will go away" and that's never going to happen.

GJC

Re:GNU/Linux as a High-End competitor...

[Anonymous Coward]

" So long as Sun continues to ignore GNU/Linux as a serious high-end competitor to Solaris, GNU/Linux has an advantage."

Yeah, 'cos Linux works really well on 100+ way CPU servers, doesn't it...

Re:GNU/Linux as a High-End competitor...

[jdgeorge]

Sun seems to be of the opinion, "If we ignore this GNU/Linux thing it will go away" and that's never going to happen.

No, you don't understand. Sun thinks that if it invests heavily in "enhancing their Unix IP license" in order to support the SCO PR/legal war against Linux, it can make this GNU/Linux thing go away.

Re:GNU/Linux as a High-End competitor...

[borgheron]

Please be so kind as to point me to the specific statistics you're referring to which prove that Solaris can handle 100+ way servers.

Besides, doesn't it start hitting the law of diminishing returns after a certain point since the amount of processing needed to keep that many CPUs straight is non-trivial?

GJC

Re:GNU/Linux as a High-End competitor...

[borgheron]

This is probably what they're thinking. I would love to see GNU/Linux crush Solaris and Sun and all their hopes.

GJC

Re:GNU/Linux as a High-End competitor...

[Anonymous Coward]

Are you seriously claiming Solaris does NOT support 100-way + servers?

Interesting...

Re:GNU/Linux as a High-End competitor...

[Anonymous Coward]

Well, if you had RTFA you would have seen that they are taking
Linux very seriously. In fact from my reading of the article
I got the impression they are impressed with how far along
linux has come and are beefing up Solaris to stay ahead.
A little competition rarely hurts.

Re:GNU/Linux as a High-End competitor...

[TheLinuxWarrior]

You obviously don't have any sort of relationship with Sun.

I work for a global pharma company as a Solaris admin and we have a dedicated Sun account team. From my discussions with our reps, I can tell you that Sun takes Linux quite seriously.

We were actually considering buying an HPC cluster of Linux X86 blade servers from Sun that their own Professional Services group would support.

If that doesn't say they take Linux seriously, I don't know what would.

I think the perception that Sun has it out for Linux comes primarily from McNeely's big mouth, which has gotten Sun in trouble for years. If they could silence McNeely, they'd be a lot better off all around.

Re:GNU/Linux as a High-End competitor...

[44BSD]

100 way? Hmmmm....they can do 64, last I cared, so maybe it's 100 octal.

Yet another Solaris distribution...

[cshuttle]

...without an included LVM. I'll stick with AIX, I think.

Re:Yet another Solaris distribution...

[haggisman]

Unless they removed it after Solaris 9, I think Volume Manager (ex Disksuite) does this. Yes, it does soft partitions too, in case you're about to ask :)

Re:Yet another Solaris distribution...

[DrMindWarp]

You mean like Sun Volume Manager as bundled with Solaris 9 ?

Re:Yet another Solaris distribution...

[cshuttle]

No, I mean an LVM that allows me to extend FS's on the fly, add in disks dynamically, and allow me to create partitions without having to remember how many cylinders I have remaining on the disk.

Solaris may be many things, but in this area, it is still older than old by comparision to HP, AIX and even RedHat AS3, which now DOES have an LVM included. Get with the program.

Re:Yet another Solaris distribution...

[Anonymous Coward]

So you mean you haven't heard of Solaris Volume Manager?
The one that has come with the os for over 4 years now...
Of course not, because your simply a troll.

Re:Yet another Solaris distribution...

[Anonymous Coward]

extend FS's on the fly, add in disks dynamically, and allow me to create partitions without having to remember how many cylinders I have remaining on the disk

Umm.... yep, yep, and yep. For that matter, you could extend FSs on the fly in DiskSuite, you've been able to add disks dynamically for about 15 years. Soft partitions (new to SVM) allow you to create as many partitions as you want at any time.

Re:Yet another Solaris distribution...

[DrMindWarp]

Ok. How about Sun Volume Manager as bundled with Solaris 9 ?

Re:Yet another Solaris distribution...

[cshuttle]

Effective, but not even remotely elegant. Use AIX's design for 1 week, and I guarantee your life will never be the same.

Re:ibm,hp,compaq *don't* make high end x86 servers

[Anonymous Coward]

Err, IBM makes 32-CPU POWER4 systems and soon 128 CPU POWER5s, HP make 128 CPU IA64 systems that would munch that thing.

Err, what do you propose Intel do with that stupid cache coherency? Just allow the CPU to use invalid data, eh?

Newsflash, one of the two axioms of SMP scalability is to minimise multiple CPUs touching the same data. Guess why? I assume you think sun sparcs do something better? Enlighten us please.

a bit of detail, and what they dont tell you

[RouterSlayer]

first a correction.
the numbering scheme, what people here have said is a bit off.

from SunOS 4, it went to 5, 6, 7, 8 and finally 9, but these are not actually v8, v9, etc as such. You see, v8 is actually 2.8, and v9 is actually 2.9
so the old question still begs, will it be 2.10 or will it be 3 ? inquiring minds want to know. (just a bit of fun here really).

and on other news, the stuff that Sun isn't saying (at least, not publically), and this is just somewhat rumor based on what I know from the inside...

Supposedly Solaris10 will be using a totally new kernel, which will be (as far as I've been told by very high ups at Sun) something of a mish-mash of Linux kernel, BSD, and Solaris.

Sun thinks it can finally "level" the playing field, and make "Unix" a standard, one flavor. All for one and Sun for all! (heh)

Many admins I've talked to say this is a great idea. Finally one standardized "flavor" of Unix, one set that consolidates everything. It could work, but is Sun the one to pull it off, and is the timing right?

We all know Sparc is dead, Sun said so themselves. So, now that they are going AMD, their direct "competitor" (I don't like using that word, because it's really not true) is Linux (and BSD of course). And we all have been hearing that BSD is dead. From a CLI point of view Solaris most resembles BSD, but from an actual operations point of view, kernel, threading, etc Solaris most resemebles Linux, after all, it already uses Linux's threading model (and much, much more!).

Word on the street from inside Sun VPs is that Solaris 10 will have (primarily) the Linux kernel. This could be amusing. I sure hope it works out, I actually like Solaris as an OS an aweful lot, it doesn't have the zealot, uptight, holier-than-thou BSD attitude... which is a good thing...

Re:a bit of detail, and what they dont tell you

[Anonymous Coward]

the numbering scheme, what people here have said is a bit off

While others may have been off, I believe you are also a bit off. SunOS 4.x became SunOS 5.X when they switched to SVR-based. Solaris 2.X was SunOS5.X+OpenWindows until Solaris 2.6. After 2.6, they dropped the 2 part (marketing hype to keep up with everyone who quit using minor numbers), so Solaris 7 is SunOS 5.7, 8 is 5.8, and 9 is 5.9. There has never been a 2.7, 2.8 or 2.9, so there's no question of 3. At best you might question whether Solaris 10 will be SunOS 6; but there's never been any comment or rumor about that.

a mish-mash of Linux kernel, BSD, and Solaris

That rumor has never been mentioned in any reasonable circle. Yes, the kernel is significantly changed (I wouldn't call it a whole new kernel), and yes there are BSD and Linux based features in the OS (probably not the kernel, though).

We all know Sparc is dead, Sun said so themselves

Would you like to provide a cite? To paraphase Mark Twain, "the rumors of SPARC's death have been greatly exaggerated". Considering that Sun just announced the UltraSPARC III+ (90nm technology), the UltraSPARC IV (dual-core or throughput computing), and the UltraSPARC IV+ (again, 90nm) and have a published roadmap at least 5 years out, I don't see where Sun has pronounced Sparc dead.

Solaris 10 will have (primarily) the Linux kernel

Have you actually downloaded the preview releases from Solaris Express? What could possible make you think that the kernel is essentially the linux kernel?

Re:a bit of detail, and what they dont tell you

[Anonymous Coward]

> Word on the street from inside Sun VPs is that Solaris 10 will have (primarily) the Linux kernel.

Well, that VP should be fired then because this is utter bull****... I think you have been taken for a ride :-) (working for Sun, BTW)

Re:a bit of detail, and what they dont tell you

[cpghost]

Supposedly Solaris10 will be using a totally new kernel [...] something of a mish-mash of Linux kernel, BSD, and Solaris.

While BSD folks won't object if Solaris 10 contained BSD code (all previous Solaris and SunOS releases did), Linux folks will have to enforce compliance to the GPL if Solaris used parts of the Linux kernel. If the rumors were true, Solaris kernel will be GPLed, and we'll soon be able to look at their sources! Great!

Re:a bit of detail, and what they dont tell you

[Anonymous Coward]

and i thought everyone new the Solaris was still on 2.x, but the marketing guys like to leave off the 2. Since when did Sun say sparc is dead? They just came out with Sparc 4 and have plans for Sparc 5. First off, a motherboard capable of 10 CPU's is not available commercially. You have to buy the whole system from IBM, HP, and NEC to get that kind of SMP system. PCI-Express will still suck for applications that really need multiple CPU's. In those areas, Solaris hardware still rocks and kicks x86 ass. Have you ever tried to switch out a CPU, RAM and Harddrive on a x86 box without shutting down the system? People do it on a solaris box without any problems. Well assuming the person doing it is trained and skilled.

I use linux, but you're smoking crack if you think cheap hardware has the same level of reliability and flexibility as high end IBM and Sun hardware. They are expensive and most people can't afford it. Those who can, the power to keep the systems running during repair is crucial.

Re:a bit of detail, and what they dont tell you

[Anonymous Coward]

We all know SPARC is dead ??? really ...

Sun's current road map for SPARC shows it splitting into 2 distinct roles (as apposed to the current i/s/e suffix Sun gives the CPUs right now.

The 'Data Facing' chips will continue to be evolutions on the current SPARC theme. US IV (Jaguar) has just been released and following that US IV+ (Panther) will be ready to roll. US V (Millenium) will be a major revision of the architecture but will continue with the same design philosophy and will be targetted at the high end, data facing applications.

In tandem with that there will be the 'Network Facing' chips. These are the Throughput Computing CMT CPUs Sun has been talking about. Gemini (2 x redesigned USIIi (yes, UltraSPARC 2i) cores on one chip) will be out in the next 12-18 months followed by Niagra (a totally new design, contributed to by Afara).

The recently released US IV dual core CPU seems to be a bit of a mix between the two, Sun are pushing it as their first throughput computing chip, but the first real 'network facing' CMT chip will be Gemini within the next 12-18 months.

Re:a bit of detail, and what they dont tell you

[demon]

At best you might question whether Solaris 10 will be SunOS 6; but there's never been any comment or rumor about that.

I've been told they might have too - too many thing might think that SunOS 5.10 == SunOS 5.1, and that might cause an assortment of nasty problems, so they may have to call it SunOS 6.0. Apparently they even have an internal list of "Things that Don't Like SunOS 5.10"...

Re:a bit of detail, and what they dont tell you

[donuthole]

Oh my god, I don't know what you're smoking - but can I have some? First off, minor nitpick. v8 is not 2.8, it's 5.8. and v9 is not 2.9, it's 5.9. Anyway, that's pointless. Solaris 10 is DEFINITELY not using a "mish-mash" of Linux kernel, BSD, and Solaris. Think about the GPL for a second, and realise why what you said is completely absurd. We have kept the same kernel, and have added so much more to it in terms of the new features - but like every other Solaris release, they are based off the pre-existing kernel. I don't know who's been telling you they are a VP, or a "high up at Sun", but I would check their credentials. -steve (who works in the Solaris Kernel test team, and gets to play with the Solaris kernel every day)

Re:ibm,hp,compaq *don't* make high end x86 servers

[Anonymous Coward]

Err, what do you propose Intel do with that stupid cache coherency? Just allow the CPU to use invalid data, eh?

No, the Intel architecture just lacks the special-purpose hardware and memory bandwidth to maintain cache coherency anywhere near as fast as the SPARC architecture.

Newsflash, one of the two axioms of SMP scalability is to minimise multiple CPUs touching the same data. Guess why? I assume you think sun sparcs do something better? Enlighten us please.

That's nice if you can find programmers smart enough to do that. It's cheaper, faster, and more reliable to throw hardware at a problem.

There are certainly things the Intel architecture does better than SPARC, and Sun is certainly feeling the pressure from it. But right now there are still things you can only get from Sun (unless you want to risk SGI being around in five years...)

Solaris 10

[Anonymous Coward]

Time to dust off the old Johnny Mathis albums and try to locate the song "Too Much, Too Little, Too Late".

Linux TCP/IP Performance

[Anonymous Coward]

"One of the little secrets of networking is high speed interfaces can in fact pump lots of bits, but they chew up lots of CPU, which means you aren't doing other things. We worked hard on efficiency, and we now measure, at a given network workload on identical x86 hardware, we use 30 percent less CPU than Linux."
from - http://www.theregister.co.uk/content/61/33440.html

Great, but what version of Linux? Has Linux 2.6 made improvements against 2.4 in TCP/IP performance? I've long understood that Linux compared to current Solaris is slightly faster in this area, and being surpassed while embarassing may highlight weaknesses most of us never encounter. I mean, how many 100 gigabit networks are there out there? But as with all things they will one day be prevaliant. Shall this be a Linux 2.7 issue?

Sol10 on other Hardware

[Anonymous Coward]

So will there be some kind of trick that I will only run on Sun hardware or could I drop it onto something else with an AMD64 proc?

Which is better?

[Anonymous Coward]

A. A talking frog
B. Sex with a mare

Waking up and smelling the java

[Anonymous Coward]

Unix will be back. Really, it will. Customers will return to Solaris one day! After all, if Schwartz said it, it must be true.

Schwartz, however, sees the fad of Linux wearing off in big businesses.

"There will be a transition back to Solaris," he said

and even Scott is a believer:

The "fad will wear off, and big business will come back to solaris".

Sun, don't worry, everything is great. Everybody else should wake up and smell the java

And I'll trust an enterprise deployment to a company with individual leaders with the brains to make the above statements on the record.

Re:Sounds like OS X!

[Anonymous Coward]

Hmmmm Solaris 10..... Apple OS X (10 in roman),
I wonder if Sun called Apple for some help here!
(haha).

Anyways, it's just a matter of time til linux
get LPARS or partitioning builtin to the OS.
As for speed, Solaris probably borrowed alot from
Linux and compiled agains't their compiler vs. Gcc.

Will Solaris be free? I doubt it! Thus, go Linux!

So...

[OneFix at Work]

Am I likely to upgrade from Solaris 9 (or even Solaris 8) just so I can gain a few clock cycles on I/O performance??? Not likely.

Am I going to risk damaging my user data, lose email, break automated processes and custom applications just to run the newest version of Solaris...not likely...

In a shocking move..

[joshua404]

Sun will be shifting the core of their OS X from Unix to MacOS System 7.

A few wishes for Solaris 10

[cpghost]

It would be great, if Solaris 10 came with a preinstalled, recent GNU environment suite (gcc, binutils, autoconf, GNUmake and the whole enchilada). It is painful to have to add this from backups or external web sites/CDs. Sun, please put this on the installation CDs as [optional] packages.

Another wish for Solaris 10 would be some kind of port (portage) system like in the BSDs or Gentoo. If Sun maintained such a port collection, it would be a great incentive to use Solaris not only on big servers, but also on desktop machines.

Re:A few wishes for Solaris 10

[donuthole]

The companion CD (which you get with Solaris 10, or you can download as a real easy 1-file ISO image) contains all the packages for the GNU build environment. It's really easy to install, and is definitely worth it.

I believe the reason we don't put it on the installation CD is probably due to licensing issues or something. I dunno, I'm a coder, not a legal guy - but really, try out the companion CD, it's got some good stuff. :)

So...

[multipartmixed]

...was 2.5.1 odd or even?

'cause lemme tell ya, it was leaps and bounds better than 2.4.

Please...

[multipartmixed]

..stop smoking crack before posting to slashdot!

Thank you.

chroot?

[Anonymous Coward]

For example if you have a Apache webserver and it gets comprimised, a hacker can't use Apache's security level to give him elevated permissions to control another part of the OS. In a regular OS you have to allow the Apache some root control over the computer to have it work properly and a hacker can use this to violate your computer.

Isn't this why you setup chroot'ed jails?

What's the question?

[fm6]

You're assuming that "What is Unix?" is a technical question. Oddly enough, that's not true, and never has been.

Something is "Unix" if the holder of the trademark says it is. When Unix first went commercial in the early 80s, AT&T defined "Unix" as being any Unix-like OS that ran on their hardware. Nobody else was allowed to call their OS "Unix", even if the OS was simply a port of AT&T Unix to another processor. That's why we have Solaris, IRIX, and HP-UX instead of Sun Unix, SGI Unix, and HP Unix.

Nowadays the criteria is a little more logical, but still mostly legal. The "Unix" trademark belongs to The Open Group, and to get a license to use this trademark, you just need to prove that your OS fully implements the Single Unix Specification. You don't even have to have based your source code on an existing Unix implementation.

But there's a catch: how do you prove that you've implemented the SUS? The only way is to run a bunch of expensive compatibility tests. Which you probably won't pass on the first try, because there are a lot of nit-picky little things you have to implement exactly. Some of the OS developers I've worked with at SGI and elsewhere think the necessary changes are very lame, and, if they had a choice, would avoid the whole process.

Naturally, they're overruled by upper management, and Sun, SGI, and HP all now have the right to use the Unix trademark. But they don't really use this right, except for a few references in marketing collatoral. ("Solaris is the leading Unix platform!") Linux could claim to be "real" Unix, if somebody wanted to spend the money to prove compliance with the SUS. But nobody does, because Linux hasn't needed this to be widely accepted.

The same sobbery is getting old

[Anonymous Coward]

Get your nose down. Have you ever worked in a mission-critical environment? I love how the Sun fanboys always resort to the same garbage: "we're talking about really important production servers, not your little FTP site or web server".

It just so happens that plenty of web farms lose money with downtime, and I would argue that you can pull and even trash a machine in a production environment if need be. Ever use WebSphere Appserver from the 5 series?

Do some homework on the industry that's flying past you and we'll see you when you catch up.

SunOS X ?

[BestNicksRTaken]

Oh come on, they just have to! ;o)

Re:SunOS X ?

[demon]

Actually, a friend of mine who works for Sun was joking about them using that moniker awhile back, but apparently they decided against it. Darn.

Solaris A

[MrChuck]

ok, the marketoids created the "Solaris (1|2).x" stuff.
Then then didn't like that "2." part, so they gave it up. "Solaris 8" preceded "Solaris 9"

What comes after 9?
A

Pretty clear. Of course if Sun weren't a company driven by technology and were driven by marketing fluff (and 1.2GHz Ultra 3's delivered a year + late wouldn't indicate that so much, esp when 3+Ghz intel boxes are CPU faster for 1/8th the cost :), then perhaps they'd use a decimal system.

"Underpinnings"

[MrChuck]

What is with this word?

It's an apple phrase that they came up with to push that OS X was aware of BSD. Too vague to really nail down, it works for apple because they're kernel is not BSD. Their utilities are.

Calling Mac OS X a BSD is like saying that Linux is a GNU OS. It's just utilities.

Now, would someone port BSD's userland over to a linux kernel and get RMS to STFU?

--
There are 10 kinds of people
Those that understand binary
and those that don't

Re:my experience with slash-dot

[Anonymous Coward]

Very amusing! Thanks for the laugh. As a 'technical IT professional' and not a 'geek' or 'nerd' I wholeheartedly concur with the very valid points you so humourously trolled there.

Well done, sir/madam.

you are correct, I was not clear....

[Desmoden]

What I was trying to say is that odd releases are often when new features/ideas are added. Why I listed them as test/play releases. And then the next even release is when more people jump on board and support/take advantage of it.

but in my limited experience I have never been able to do a full site upgrade to the odd version. I've usually had to wait for the even release for everyone to "bless" their code/products.

which has never bothered me as it's a lot of work for me as well. So I prefer to only upgrade everyone on the even releases.

Don't get me wrong, I have systems running 5.9, but nothing with critial apps.

Re:Sun OS X

[Anonymous Coward]

dumbass moderators don't know sarcasm.

Re:Not just Solaris

[Anonymous Coward]

hmm, I've run debian on my powerbook for going on 4 years now.

Never had a problem compiling software, but I rarely have to bother, since the archive is so complete.

Re:True, True

[Bastian]

I work on a couple x86 linux boxen, a Sun workstation, and my Powerbook running OS X. I frequently write programs that will run on the big-endian machines but not the others and vice-versa. Sometimes it's just a pain in the ass to handle endianness issues, especially if you don't personally plan on running your software in Lilliput or whatever.

I'm wondering if maybe it would be possible to write some functionality into gcc or maybe libc for handling general endianness issues. I would be a happy boy if I could set a flag and have all the byte-flipping handled for me. That said, I wouldn't want to be the poor sap who has to try writing a hack like that and at the same time designing it to be usable.