Slashdot Mirror
Sun to Make Solaris More Linux Like
ramboando writes "In an effort to spur adoption of Solaris, Sun Microsystems has begun a project code-named Indiana to try to give its operating system some of Linux's success.
Sun has been trying for years to restore the luster of Solaris, but that since has faced a strong challenge chiefly from Linux. Sun wants to embrace some Linux elements so "we make Solaris a better Linux than Linux," said Ian Murdock, Sun's chief operating systems officer, quoting Netscape co-founder Marc Andreessen, whose latest start-up, Ning, uses Solaris.
But it's a tricky balance to adopt elements of Linux while preserving Solaris technology and advantages such as the promise of backward compatibility. "As we make Solaris more familiar to Linux users, we don't [want to] lose what makes it more compelling and competitive.""
I've liked many aspects of Solaris for a long time, but the #1 thing that turns me off it is the userland tools.
Yes, I know they ship a DVD with lots of GNU tools, but the fact that the built-in make, vi, grep, etc. are still basically unmodified from the early 1990s (if not longer) is not, to me, a feature. Those hoary old versions should be the ones on a supplementary DVD for those who need perfect backward compatibility with 15-year-old shell scripts and so forth.
It sounds like that's a focus of this project, so I say fabulous. If I can get ZFS and DTrace plus a modern toolset out of the box, Solaris will start to look much more attractive.
ZFS? DTrace? Zones?
Sun is a for-profit entity. How do they expect to make money off of their OS? They should GPL Solaris, let the code monkeys snatch the best bits for Linux, and forget about wasting their money developing Solaris. They can write a "shim layer" on Linux for people needing backward compatibility so they don't alienate long-time customers. They need to figure out where they plan on making money, and scrap the parts that lose money. Open sourcing Java was an indication of desperation; we saw plenty of companies open source their product during the dot-com bust, either because they didn't want their work to die, or because they thought it would magically boost market share and generate revenue. It doesn't.
Solaris is open source and free. http://www.opensolaris.org/os/
Also consider that some of the better solaris features have been added to FreeBSD recently. dtrace and zfs are available for FreeBSD 7 current.
MidnightBSD: The BSD for Everyone
I'm currently struggling to implement a Solaris server right now. The user space is archaic, obscure, and seems to be difficult for the sake of being difficult. Things like updates are still done the way they were done 15 years ago, often requiring a drop to single user mode (as bad as a reboot in my opinion), and often require a system reconfigure. Solaris' kernel is cutting edge and, in some ways, way ahead of Linux. But in the ways that count, Solaris lags far behind.
/usr/bin, usr/local/bin, /opt/sfw/bin, /usr/sfw/bin, /usr/ucp, etc. I frequently find that I have to compile things from source just to get basic functionality. For example, Sun ships Samba with solaris, but it doesn't support LDAP. They also ship some hacked kerberos libraries, based on MIT, but if you need to build anything that depends on kerberos, you have to compile and install a separate set of MIT Kerberos libraries. Some apps are available in package form (solaris packages) from sunfreeware.com that you can pkgadd. But PKGs don't seem to be a complete packaging system like deb or rpm is. The pkg-get utility from the aforementioned site is very useful, though.
Just to make the system usable requires a ton of third-party software that sun does not ship nor support. In the end my path has nearly half a dozen bin folders in it, by the time you could
The init system is currently in a disorganized state. Most things are migrating to svcadm, which under the hood is very much like launchd. But there are still init.d scripts, but they don't always work right. Maybe Linux should move away from init.d, but at least on redhat, they are very full-featured and quite easy to work with.
Sun's biggest strengths right now are zones, zfs, and dtrace. However, if you don't specifically need these features, Linux is a better choice in many circumstances. And Linux is gaining features in these areas. xen can do a lot of what zones do, albeit much less efficiently. dtrace functionality is coming, I hear. ZFS, well the kernel developers seem to be suffering a bad case of NIH syndrome. The only reason I'm using solaris right now is ZFS. But I'm taking a big risk deploying it on a 12 TB disk. I have yet to hear of a failure, and Sun assures me that it's enterprise-ready. Sun's assurances do carry a lot of weight; they've had a lot of experience in these things. But I'm only a silver-level support customer. It's taken two weeks and some 20 phone calls to get issues sorted out with our sunsolve account and updatemanager. Our assigned support group only wants to talk over e-mail, which is annoying. Turnaround time on trying out their suggestions is hours if not days. This certainly isn't quite the same Sun as in the olden days.
Anyway, talk to any Sun jocky and he'll tell you that none of my complaints about Solaris are weaknesses. They are strengths. Cryptic commands are second nature. Besides, they separate the real sysadmins from the wannabes. Sound familiar? I think I've talked the same way about Linux to my Windows friends. I'm glad that Ian is going to work to improve Solaris' user space (which is what he means when he says make Solaris more like Linux, right?). On the other hand, Solaris reminds me not to get complacent with the state of linux. Every complaint I have about Solaris could easily be echoed by a Windows refugee trying to make sense of Linux. Both Linux and Solaris are powerful, cryptic, and archaic OSes. They both have a lot of room for improvement. We'll have to see. I told my RedHat friend the other day that his company has nothing to worry about from Solaris. Hopefully Ian will change that.
Anyone who has managed very high load webservers already knows that solaris has significant advantages. a much better effort would be a grass-roots effort to educate the Linux community of why 10+ years of professional development lead to significant performance benefits on multi-core, multi-processor systems.
Solaris serves a niche in the market that is growing like crazy now, and most web developers who are building apps today should look into it seriously, IMHO.
May I add: Fault Management Framework [1], Crossbow [2], pNFS [3], stable device driver interface (one of the biggest point driver developers complain about in Linux). Clearly the GP has no idea about the number of technological advances Sun is pushing in OpenSolaris.
[1] http://www.opensolaris.org/os/community/fm[2] http://www.opensolaris.org/os/project/crossbow
[3] http://www.opensolaris.org/os/project/nfsv41/pnfs
Excuse me?
/home and /usr/local, and a few other directories, and copied the relevant entries from /etc/passwd and /etc/group. Copied whole applications, their environments, etc.
I just migrated an entire system from Solaris 7 to a Solaris 10 Zone - How? I tarred up
Solaris 7 is from 1999, and this is 2007. Try that on an 8 year old redhat box and see what happens. Good luck with that.
...and all the functionality of Linux has jumped ahead of Solaris... arguing that Linux may eventually catch up with these powerful Solaris features is a little disingenuous don't you think? Linux and Solaris are both worth having, depending on what you need. I look forward to what this project, and the OpenSolaris project, can put together.Craft Beer Programming T-shirts
I've been using sunos and Solaris and sun hardware since '86. I can build a very security solaris 9 server that ends up with about 5 packages and a few things from a few other packages so it results in a nice simple stripped down system that is just enough to run the application and its great for systems that live in data centers.
Then sun comes along with Solaris 10 and adds in a ton of complexity with out providing any additional services. The new things like zones and zfs don't need all the new extra crud but its nearly impossible to build a lean system with solaris 10. There are also a number of issues that are just plain wrong and reeks of security the Microsoft way. Why does live update look inside zones? If its in a zone, its not to be trusted outside the zone. Thats covered in Security layers 101 so back to school guys. (you can purge one file inside a zone that breaks doing patches in the global zone). The new admin tools remove the rc scripts... except that most of them are just moved and hidden by layers of config files. Then it uses a binary file to figure out what to run at shutdown, and it keeps changing the file when servers start and stop and you can't get an accurate picture of the data its going to use when it shuts down the system. Since the file is a binary file, you can't checksum it and you can't dump it so you've got no clue if someone has put a Trojan in it. The data in the file could have just gone in a nice plane text file but I guess the coders missed the Windows registry too much. The appear to be handing the keys to the source castle to any old hack. Someone "fixed" telnetd and added a new feature in one of the worst security lapses I've seen in a long time.
I just bought 3 new netra 210 because 1) they run SPACR Solaris 9, 2) they fit in my racks and 3) are one RU. I'll stop buying Sun hardware the day I can't run Solaris 9 because there is no way I'm putting Sol 10 on a production machine.
Not necessarily a bad thing, vegemite is much more palatable than caviar
I'm using Solaris because the data mining application I'm building (in Lisp) brings the Linux kernel absolutely to its knees. Solaris runs it just fine on the same hardware. (We're talking 30+ GB of heap -- Linux is dead meat after 3 to 4 GB.)
A friend of mine says this is because the Linux kernel hackers optimize for the common case, not for extreme cases. I suspect this is correct. To put it another way, they are more into cycle shaving than analyzing the time and space complexity of their algorithms -- just as one might expect from smart hackers with a relatively weak computer science background.
The result is a kernel that does great on normal workloads, but just falls over when subjected to unusual stresses. Unless and until this is corrected, there will be a need for Solaris.
Your god may be dead, but mine aren't!
BrandZ isn't for Linux... it is Linux, running in a Solaris Zone.
If you want a GNU-like system for Solaris, try out Nexenta
"As we make Solaris more familiar to Linux users, we don't [want to] lose what makes it more compelling and competitive."
If Solaris was compelling and competitive, they wouldn't be trying to make it more like Linux.
Solaris is something that we use as a legacy OS where I work. We have well over 700 Linux systems in the school of engineering. At last count we had maybe 35 systems running Solaris still lingering here and there in places where they either cannot be replaced or there is no economy in doing so. There has not been a NEW installation of Solaris deployed in at least two years. We've also got five Tru64 systems, two HP-UX systems, three Irix systems, and I think 4 VMS systems that a dedicated die-hard won't allow to expire.
The bottom line is that the unix wars are over, Linux has won, and whatever contender eventually does take the crown from it will NOT be one of the has-eens of the past.
I'm long past caring what Sun does or does not do with Solaris for the same reason that I don't care what E-com does with OS/2. Both OS's may or may not be configured with fancy new features in the future, but it doesn't matter because they've already lost.
Game over dude, and no you don't get your quarter back.
Muslim community leaders warn of backlash from tomorrow morning's terrorist attack.
You should look at you kernel parameters ulimit -a As shipped Solaris is intended for big iron in a way that most Linux distros aren't
"Linux is for noobs"-The new MS fud strategy
Coming from a BSD background, the thing I dislike the most about Solaris is that it refuses to have a 'minimal install' that is actually usable. I can install *BSD in a few tens of MBs, and then add the packages I want easily. This makes it easy to secure and run the machine, because I know exactly what's on it. Last time I installed Solaris, the base system seemed to be about 4GB.
I am TheRaven on Soylent News
Caseih" is correct when he says "This certainly isn't quite the same Sun as in the olden days", in regard to support and how it is delivered. It certainly isn't the same Sun for those of us who are tasked with delivering support. Management has implemented all sorts of programs to improve customer "sat" and bring down call hold times, programs that INTERFERE with the day to day support work; effective and seasoned TSEs are bailing out right and left and ARE NOT BEING REPLACED in many cases; the EDS "partners" have a large turnover rate (what do you want for $9 an hour?); more time on the phone taking live calls, meaning the TSE have less (or no) time to do followups, research, spend time in the lab . .
The "Dell-ization" of tech support is spreading like a virus; support is a commodity now. Even enterprise level tech support. Sold to the lowest bidder. Who cares if the person on the phone can't spell "LDAP", as long as the call is picked up in X minutes and keeps the manager's pager from going off? THAT is where Sun support is today.