Sun Solaris Vs Linux: The x86 Smack-down

JigSaw writes "Tony Bourke put together a long article, benchmarking File System, System, Compilation, OpenSSL and Web Performance for both Linux and Solaris on x86 hardware. While SPARC's Solaris is said to be more optimized than its x86 counterpart on the other hand so is Linux 2.6 compared to 2.4. Solaris-x86 performed well in the tests, but Linux 2.4 seems to win most of the tests and the overall impressions."

Comments

Sorry?

[Sir Haxalot]

While SPARC's Solaris is said to be more optimized than its x86 counterpart on the other hand so is Linux 2.6 compared to 2.4
Maybe for computers with multiple processors, for regular computers Linux 2.6 is comparitively slower.

Re:Sorry?

[toddler99]

whatever solaris blows linux rules nuff said

Re:Sorry?

[Anonymous Coward]

you speak absolute rubbish!

my 2.6 box flies along compared to 2.4

sure, long running tasks/crunching is slightly
slower, but overall responsiveness is so much
quicker. I'll never go back to 2.6 on Linux
(On Solaris I'll stay as far away from 2.6 as I can! ;-) )

Hardly definitive

[burgburgburg]

I see nothing in the article about a steel cage. You call that a smack-down?

Huzzah!

[gazbo]

Proof positive that Linux is faster than a famously half-arsed unoptimisedport of an operating system! I'll break out the Champagne; for a victory of this magnitude we'll need something extremely expensive - like Cava.

Re:Huzzah!

[Anonymous Coward]

Linux is Cava...you oaf.

Huh, because both are unoptimized?

[bconway]

While SPARC's Solaris is said to be more optimized than its x86 counterpart on the other hand so is Linux 2.6 compared to 2.4.

Because both are unoptimized, these are suddenly comparable differences? You're comparing apples and oranges, and quite frankly, it's laughable. Why would you think the two are in any way related or would yield similar differences in benchmarking?

Re:Huh, because both are unoptimized?

[Anonymous Coward]

No kidding. The fact that Solaris beat Linux on x86 in as many benchmarks as it did is downright scary. Linux can't even be faster than Sun's afterthought? How is it ever going to put BSD in the grave?

Re:Huh, because both are unoptimized?

[Otter]

You're comparing apples and oranges, and quite frankly, it's laughable.

In fairness, the article itself compares apples to apples. It's the submitter who decided to add, "Of course, oranges are more orange than apples but so are peaches!"

Re:Huh, because both are unoptimized?

[Frymaster]

in real fairness... the benchmarks are not what they should be!

1. i assume he used the tar that ships with solaris... and we all know that the solaris tar sucks badly. if he wants to benchmark the fs and not the app (tar) he should have installed gtar on the solaris box.
2. the http benchmarks seems specious... i would have preffered to have seen a specweb99 bench... something that i could compare to real, existing benchmarks.

of course, having said that, i'm still not going to install solaris on my x-whatever-6 box any time soon. no. not even for openwindows

Re:Huh, because both are unoptimized?

[Ed Avis]

The biggest gap in the benchmarks (IMHO) is the gcc comparison - it's well known that gcc-3.2 is much slower than 2.95. He should have used the same gcc version on both OSes, since this was an OS benchmark not a gcc benchmark.

Re:Huh, because both are unoptimized?

[pebs]

The fact that Solaris beat Linux on x86 in as many benchmarks as it did is downright scary.

I don't know if I'd say its scary. Take a closer look at the benchmarks. The GCC benchmark compares 2.95 on Solaris vs 3.3.1 on Linux, making it pretty useless as a benchmark. Why didn't they compare to 2.95 on Linux? 2.95 compiles faster than 3.3.1 in general, from what I hear.

The MySQL benchmarks were mixed results, with inserts being one of the areas where Solaris was faster. And we have only compared Ext3 here, which is one of the slower filesystems for Linux (because it journals data). Everywhere other benchmark, Linux was faster.

Sure, Linux has much room for improvement. Is there anything to worry about because of these somewhat flaky benchmarks? I think not.

Re:Huh, because both are unoptimized?

[Anonymous Coward]

> You're comparing apples and oranges,

Huh? Mac wasnt even in the article!

Talk about RTFA...

Re:Huh, because both are unoptimized?

[hackstraw]

Because both are unoptimized, these are suddenly comparable differences?

Yes, both are what you would get "out of the box".

Re:Huh, because both are unoptimized?

[drinkypoo]

He just mentioned it in passing. The reason he is comparing these two is not that they are both unoptimized, but that they are both standards. It's well known that Sun's "professional" compiler suite produces some of the best code around for assorted sparc-architecture processors, which makes sense on assorted obvious levels, but their lack of fine optimization on x86 may harm them, and this is (if amazingly vaguely) alluded to here.

Meanwhile Solaris 9 for x86 (aka SunOS 5.9, as the article says - it misses the point that Solaris simply means (in Sun nomenclature) SunOS plus the windowing environment, and it once means SunOS plus openwindows, and that Solaris 1.x is SunOS 4 (BSD-based, mentioned) and Solaris 2.x is SunOS 5 (SVR4--based, which is not mentioned directly that I recall)... Er, where the hell was I? Solaris 9 is not available in a version well-optimized for x86. Because you can only relink the kernel and not recompile it, since source code is not provided, it is doomed to this fate. Redhat 9 is also something of a standard, and it happens to come with (and only support) linux kernel 2.4.20. 2.6 has many optimizations but is new. So he mentioned it because it puts both distributions on somewhat equal footing, and in fact in most benchmarks (which are overly simple, but anyway) the systems came up with similar performance when realistic options were utilized.

Re:Huh, because both are unoptimized?

[jedidiah]

Why should most of what Solaris do be dependent on the CPU? Quite a bit of what Solaris does should actually be quite independent of the underlying hardware. If Solaris SPARC is really "all that" then even the "red headed stepchild" should be able to keep up with other operating systems.

Re:Huh, because both are unoptimized?

[jafac]

"It's well known that Sun's "professional" compiler suite produces some of the best code around for assorted sparc-architecture processors,"

I've heard this said over and over - but I once worked on a project where we hit a memory leak in our code - our developers traced it down to the memory management code in the Sun compiler. Set our development team back two years. Sure this is an isolated case, and I don't know if you can take that to mean that Sun's compiler isn't top knotch. But I'm certainly bitter about it, and I know my former team and our customers certainly are.

Re:Huh, because both are unoptimized?

[drinkypoo]

It might produce buggy code, but it's certainly fast code. The value of a compiler which allows you to get to its compilation errors more rapidly is certainly debatable. I never had any problems with it personally, but that doesn't mean a whole lot I guess, since I'm not a programmer type, just a sysadmin type.

Re:Huh, because both are unoptimized?

[melonman]

It might produce buggy code, but it's certainly fast code.

If the bug involves a memory leak, as in this case, I guess that means that your server falls over faster than would otherwise be the case.

Sun on x86

[devphaeton]

Was Sun really serious about Solaris on x86?

The fact that they simply "give away" the OS for cheap (i actually got my copy for free from Sun) kinda makes me think they've only released the x86 version to make it "available" to more poeple.

The more people that are familiar with Solaris, in theory, the more Admins/IT staff will end up recommending SUN hardware/software at their workplace. It's a marketing strategy. Not a pervasive strategy, but a strategy nonetheless.

If you take my meaning, mr. Frodo.

Re:Sun on x86

[n3rd]

Was Sun really serious about Solaris on x86?

On and off, evidently back on again. I've heard that back in the mid 90's a decent amount of customers used Solaris x86 on Compaq's. After a while they dropped support and over the next few years. Anyone confirm or deny (I know the second part is true)?

Here is a recent press release about Solaris x86. Disregard the marketing garbage, there's a lot of it.

They name a decent amount of customers, a biomedical place is one of them. Perhaps a transition from SPARC to x86 for sheer speed would be cheaping going from Solaris to Solaris instead of Solaris to Linux, that is assuming Solaris on x86 meets their needs.

Also, according to this article they have Solaris x86 for Opteron. Perhaps this would help convince big graphics apps such as Photoshop make a port to Opteron since Linux and the BSD's are already there.

They also have a POWER4+'esqe chip coming out in the first half of the new year. Two UltraSPARC III cores with 8 megs of cache and each running at 1.2 GHz each.

Sun has good things going for them but they need to expand into new areas and take another look at the current situation.

Re:Sun on x86

[devphaeton]

Thanks for the info and the links :)

Sun has good things going for them but they need to expand into new areas and take another look at the current situation.

Seriously. I still think SUN is a good company and they have the potential to rise and be great again, but they've really let themselves languish over the years. I would hate to see them (or SGI) go away. But they need to make some changes, and do it fast, since it's getting close to being too late.

Re:Sun on x86

[Anonymous Coward]

They weren't very serious about Solaris 9 on x86 -- the product wasn't even publicly released at first. Sun only made it available after an outcry from Solaris shops.

However, Sun says that it's getting more serious about Solaris x86, and that future versions will be more competitive with Linux. (Something they shoulda done 4 years ago...)

For example, the Sun 'Java' desktop runs on Linux, but future versions will run on Solaris instead.

Re:Sun on x86

[drinkypoo]

Sun sold a lot of x86 licenses to companies which wanted to keep the same OS all around their enterprise, but also wanted workstations cheaper than they would be with SPARC processors in them. Around the time of the 486, the x86 processor started to really deliver on price:performance; when the pentium came out, its future as a leader in that category was assured. Today you cannot get as much performance out of any processor for as little money as with x86-compatible processors, and the other architectures have been relegated to niche markets in the low end, and highly scalable servers with loads of processors in them, something x86 has not traditionally done (though we may be a step closer now, with AMD's Hammer architecture, and the use of Hypertransport links.) Hammer supposedly supports up to 31 CPUs - 32 devices, counting the chipset. With its 64 bit addressing, it could lead the way to so-called "Big iron" machines based on x86-compatible (though also x86-superior) processors.

Re:Sun on x86

[WoTG]

I think they are rather serious about it. But, as you say, not necessarily as a profit generator right now.

First, IT staff who try Solaris on x86 will NOT be any more likely to recommend Sun gear if Solaris x86 is intolerably slow or buggy or whatever. It doesn't need to be the fastest, but it can't be too far behind Linux and I guess Windows.

Also, x86 becomes insurance. It's tough and expensive to design competitive chips - and getting tougher all the time. If somewhere down the line Sun considers giving up on Sparc, then having a reasonably good x86 version gives them options! They can always package up x86 hardware to sell entire systems (i.e. they don't have to drop hardware sales) - yes it's would be difficult to get the same level of uptime on x86 chips, but it's also difficult to design software and hardware.

Re:Sun on x86

[WatertonMan]

Sounds like they may be trying to move to AMD. Not a bad idea although Xeons are likely as well.

Re:Sun on x86

[DShard]

I don't think anyone saw linux coming four years ago. I may have been using it since '94, but that was at a university and isolated to ultra-geekdom. Truthfully I think that it has only been taken seriously for the last two. I'm not saying that EVERYONE should have understood the implications of open source, but I still think very few people really do.

Re:Sun on x86

[digitalunity]

Hammer supposedly supports up to 31 CPUs - 32 devices, counting the chipset.
I thought I'd heard something about this. It could be that the basic Hammer architecture can support this many processors, but the latest + greatest Opteron only supports 8 processors. If the Opteron came out in a 32-way SMP capable package, AMD might finally get the lucrative big iron contracts they've been wanting for so long but lost to IBM's Power4 servers and the like. There's big money in that sector, and it looks like(right now at least) the AMD64 platform will give a price/performance metric that is unbeatable.

Re:Sun on x86

[confused one]

The >8-way processors are in the pipeline. There's support from hardware vendors who want access to it; and, AMD's promised it. Let them work out the bugs in the current implementation(S) and then they'll start producing the more advanced versions...

Why Red Hat ?

[MoonFog]

Guess that question will be asked, and to those to lazy to RTFA, here's his reply :
I chose RedHat 9 for the simple fact that it is a very popular distribution, and is ubiquitous in terms of corporate and personal deployment. Of course it is not the end-all be-all of Linux distributions, but it's both popular and effective, which makes it appropriate as an evaluation platBesides, most of what I evaluate has more to do with Linux itself, and not the distribution. The only significant affect RedHat has on this evaluation is the specific version of the kernel (2.4.20-20.9) and the use of RPMs (which some other Linux distributions use as well).form.

Re:Why Red Hat ?

[Anonymous Coward]

too lazy, not to lazy!

For ramblings on "Oracle on Solaris or Linux?"....

[tcopeland]

...from the distant past, there's this Slashdot thread from way back in 1999.

There's a "Summary of Points" post a ways down that page that nicely encapsulates most of the discussion.

yeah right

[andih8u]

If you're looking to run PHP, or Apache/Tomcat/Java (which comes pre-installed), then Solaris x86 is a solid, stable platform.

Yeah, trying to install PHP on solaris is a freakin riot. Linux wins out just based on packages alone.

Re:yeah right

[djh101010]

I'm doing that just now (installing PHP on Solaris). On the Linux box, the dependancies worked out nicely, on the Solaris 9 box, I'm in cascading-dependancy-hell. Yeah, I know, it's just making a bunch of stuff, but still...

Re:yeah right

[Anonymous Coward]

Just because you are used to compiled binaries of the RPMs does not mean that us seasoned *nix admins should be forced into such an elementary procedure. Have you even read AMPS? I could teach a 10 year old how to install PHP on Solaris with that document.

Re:yeah right

[Rasta Prefect]

Yeah, trying to install PHP on solaris is a freakin riot. Linux wins out just based on packages alone.

PHP on Solaris 2.6.1. It's a Sun 670MP, its the newest version you can run. I still have nightmares.

On the plus side, once it did work, it's worked perfectly ever since - no crashes, no downtime.

Re:yeah right

[Anonymous Coward]

AMPSWow, that looks a lot easier than say
urpmi php4

or

pkg_add -r php4

Re:yeah right

[bofkentucky]

Double check your sed, I was having trouble this summer compiling PHP4 on Solaris 8, installed the kitchen sink of gnu tools (including gsed) and things started working again. Does anyone actually use sun's tools (tar, sed, awk, vi) anymore, the gnu (or vim) versions are much better.

Re:yeah right

[kindbud]

Yeah, trying to install PHP on solaris is a freakin riot. Linux wins out just based on packages alone.

That PHP is too complex for a wanker to install without a prepared binary RPM is not the fault of Solaris or Sun, or even the PHP authors. Gee, who does that leave to blame?

Re:yeah right

[Oggust]

There's a good point in there.

One major problem I have with solaris (and I do run it on sparcs) is that the whole userland has been neglected for years. You have to install a great number of third-party apps to get a usable system. And unless you get into the solaris-package-building-business in a major way you'll do it from source, and then you have an interesting time patching all your systems when the time comes.

With Red Hat, you're pretty much set with what's included. Stuff like vim, lsof, nmap, ethereal, screen, cvs... They are all just there.

/August.

Re:yeah right

[bofkentucky]

We get most our stuff from sunfreeware, php and some other apps end up being source jobs, but I keep a list of my ./configure options for each app so configs stay the same between version switches

Re:yeah right

[Anonymous Coward]

That PHP is too complex for a wanker to install without a prepared binary RPM is not the fault of Solaris or Sun, or even the PHP authors. Gee, who does that leave to That PHP is too complex for a wanker to install without a prepared binary RPM is not the fault of Solaris or Sun, or even the PHP authors. Gee, who does that leave to blame?blame?

So, you've compiled and installed php on a Solaris box, right?

Re:yeah right

[roacheh]

And I got up at 4am to restart tomcat on linux. I think I would have prefered compiling PHP during business hours. PHP is relitavly easy to compile on Solaris for an experienced user anyway. Solaris is also the "distribution" of the SunOS kernel and tools compared to say RedHat which is the Linux kernel and tools. UFS Snapshots, RBAC, disksuite.. linux? yea right.

Re:yeah right

[bolthole]

You missed out on

pkg-get install php

http://www.blastwave.org/pkg-get

Why does this get put under developers?

[bombadillo]

In my experience the majority of Developers don't have an in depth knowledge of OS and Hardware performance. The System Administrators almost always have a much better understanding of OS performance. The majority of developers I have worked with are Java developers. Perhaps it is different for other languages. However, shouldn't this be under a Systems Admin category?

Re:Why does this get put under developers?

[MoonFog]

The System Administrators almost always have a much better understanding of OS performance.
In my experience sysadmins tend to know much about a certain OS, and will proclaim it to be the best no matter what the benchmarking results are. They may have knowledge about the OS, but rarely have any depth knowledge of the actual hardware performance.

Re:Why does this get put under developers?

[cK-Gunslinger]

In my experience the majority of Developers don't have an in depth knowledge of OS and Hardware performance.

That's true for probably 90% of developers out there, but the truly *great* programmers realize these concepts play a tremendous role in software development and try to understand them as much as possible. Any developer writing anything close to real-time software *has* to know how things are working at a very low level.

Re:Why does this get put under developers?

[awfar]

Wow, I am saddened; of course you are correct in general terms as a sysadmins lot in life means spending an *intimate* amount of time with a specific OS and don't see what the other systems can do. Some of us that have spent significant time reading, understanding, and programming the underlying OS concepts and architecture for a great many years across a great many OSs and systems (admining simply because we were GOOD at it; I started on B5000 architectures, the root firsts of many OS concepts we take for granted including virtual memory, stack-based OSs, compilers, etc.) but somehow get bundled with those who don't or didn't and don't understand what you have said. I have gotten the "Oh, you're a sysadmin... we can't allow you to do a different OS (or Programming, or design work, or whatever)" despite it was me who did the Oracle design and DBA work, did the performance analysis and benchmarks, bailed their developers out when I saw there solution was not scalable.

I urge you to seek out those who DO know the difference and give them an opportunity to assist; you will have a friend for corporate life(!) and an invaluable resource when the ball falls within their bailiwick.

Re:Why does this get put under developers?

[MoonFog]

Some of us that have spent significant time reading, understanding, and programming the underlying OS concepts and architecture for a great many years across a great many OSs and systems
I don't deny that, nor that you are good at it, I was merely saying what I have experienced. I have several friends with MSCE degrees, but that doesn't help me much when developing a program for Linux, and neither are very interested in learning about it.
I urge you to seek out those who DO know the difference and give them an opportunity to assist; you will have a friend for corporate life(!) and an invaluable resource when the ball falls within their bailiwick..
I assure you I appreciate any person with indepth knowledge who can teach me more about an OS, language etc :)

Re:Why does this get put under developers?

[Wakko Warner]

In my experience sysadmins tend to know much about a certain OS, and will proclaim it to be the best no matter what the benchmarking results are. They may have knowledge about the OS, but rarely have any depth knowledge of the actual hardware performance.

And, unless you're suggesting developers are more likely to possess this knowledge than sysadmins -- something which makes no logical sense -- you've really not answered the question.

- A.P.

Re:Why does this get put under developers?

[bloosqr]

He's right at least in the scientific community. Most of us prof/gprof our code immensely and take advantage of everything the hardware can give us at the very least on the compiler level.

This is why you'll have developers extoll the virtues of single multiply adds that IBM's power architecture has if they are doing numerically intensive code (now incorporated into the G5) or perhaps the memory management of the Sun boxes if they are running code that necessitates having > 4 gb ram.

Things like "cache size" matters to developers in a way that it does not to sysadmins. On the other hand I get the impression that much of the "sparc is god" raraing comes from development/running of Java code or running/developing websites. In any case it makes much more sense to me that developers do profiling of their own code and thus know the performance issues of the architecture than a sysadmin.

Re:Why does this get put under developers?

[g_goblin]

Well since JAVA is a "Right It Once" language, it doesn't care about the underlying OS. The runtime is supposed to be optimized for the platform it is running on.

In the case of C/C++ development, you have to know the underlying OS, especially when it comes to Services in the Windows world and Daemons in the *NIX world. The STL doesn't provide any of this for you. Threads are another issue. If you are planning on calling STL functions, you should use _beginthread, else use the functions provided by Windows or POSIX to create threads. With JAVA you don't have to worry about this. You just say give me a thread and it does.

JAVA gives you the ability to run one code base. But you are dependant upon a runtime, which can end up constricting you based upon features it supports. C and C++ do not do that. If a new feature of an OS comes out, all you need is the API to get you started.

I guess the point I am getting at is it depends on the type of application being developed and the language it is written in. A good C/C++ developer should know the internals of the underlying OS and the hardware. If they don't, then they are probably fresh out of school or writing their apps in the wrong language. Otherwise, how can you benchmark your app and make a hardware/platform recommendation or give a yearly operating cost plan?

Re:Why does this get put under developers?

[sydb]

1. MCSE's are not degrees!
2. MCSE holder's are not necessarily sysadmins!

Re:Why does this get put under developers?

[Anonymous Coward]

I can't believe what I have seen in this very thread.

Isn't it obvious that both types complement each other?!

Developers: Know their software and their tools.
Sysadmins: Know the enviroment where the software will be ran.

Simple and easy...

Re:Why does this get put under developers?

[Brandybuck]

As a systems programmer, I'm supposed to have a lot of in depth OS knowledge. Which is why I'm currently taking sysadmin classes, to get a better knowledge of it.

Every systems programmer I work with uses Linux or FreeBSD at work or at home (though maybe not as their primary syste). Most hardware engineers here do as well. But none of the application developers do. I don't know if this is the industry norm, but it is at my work.

When I have a code review, the systems programmers want to know why I used vfork() instead of execve(). The applications programmers want to know why I used a visitor pattern instead of a mediator.

Re:Why does this get put under developers?

[Spl0it]

Personally, being a novice developer myself I must disgree with you. Its not only important but vital to know what software can work with what hardware and what is the best server for an application/web site/web service I'm helping with or making on my own. I don't think a Systems Admin should be deciding what type of software should run where, although I think there input is needed and an asset.

Re:Why does this get put under developers?

[Anonymous Coward]

Don't worry. To non-java developers, java developers aren't worthy either :-)

Re:Why does this get put under developers?

[Anonymous Coward]

vfork is in no way a replacement for execve.

vfork is a replacement for fork. back before COW (copy on write) memory management schemes, fork would copy the entire address space, whereas vfork would not.

Thus you had to be exceptionally careful when using vfork (basically, normal programmers should only vfork if the very next lines are an if and an exec* in the child.

Nowadays, vfork is an incredibly marginal performance win, it's only a win at all because you don't have to copy the page frames and VM data structures (though I can't speak to whether or not modern UNIXes implement this optimization). user address space is never copied.

Frankly, vfork was a hack to workaround shitty performance in the original fork. Architectural improvements (COW addreess spaces) are always superior to hacky interface optimizations.

err umm... (i hate to say this but) duh?

[Anonymous Coward]

its long been a known fact that sun has not dedicated much in terms of resources towards creating an optmized x86 solaris; in fact just a year or so ago, they declared they may not release an sol 9 for x86 at all.

honestly, id like to see how linux stacks up vs solaris on sparc iron. personal experience easily declares ******* the winner, but then again, what do i know?

(names withheld to protect the trollish)

Re:err umm... (i hate to say this but) duh?

[SoTuA]

honestly, id like to see how linux stacks up vs solaris on sparc iron. personal experience easily declares ******* the winner, but then again, what do i know?

Well, I have some personal experience on this, and it declares ***** the winner. I won't tell you wich, but it has a cute penguin instead of a cup of coffee as it's mascot ;)

We kidnapped a Sparc workstation and put debian in it. While the installation was a "riot", the performance kicks solaris' ass into the next neighborhood. In fact, the people at Systems Support were impressed, but it seems our Sun contract forbids us from running anything but solaris on the machines :( so we just kept that token machine and the comps lab still runs slowlaris.

Re:err umm... (i hate to say this but) duh?

[Spl0it]

You make a good point, but your forgetting Linux was far from optimized in the tests for this Article, and the 2.4 kernel was used aswell.

Re:err umm... (i hate to say this but) duh?

[drsmithy]

While the installation was a "riot", the performance kicks solaris' ass into the next neighborhood.

How well did it scale ?

That's the big selling point of Solaris - it might not be fast, but it takes a hell of a lot to make it *slower*.

Re:err umm... (i hate to say this but) duh?

[SoTuA]

How well did it scale ?

That's the big selling point of Solaris - it might not be fast, but it takes a hell of a lot to make it *slower*.

Yes, I know. It probably won't scale too well on linux. But the thing is, Sun is offering workstations as a bonus for the new CS dept. server (so it undercuts Dell servers). I sure would love Sun iron on the server, but keep your Sun workstations, we can manage pretty well with dual-booting PCs.

no surprise

[Anonymous Coward]

Solaris's strength lies in scalabilitly. I have a feeling Linux/SPARC will also beat Solaris/SPARC on a single CPU machine. But keep adding CPUs and watch Solaris scale (almost) linearly!!!

Re:no surprise

[n3rd]

Solaris's strength lies in scalabilitly.

Very true, it's because damn near everything is threaded. Threading is highly encouraged by Sun when programming for Solaris on SPARC.

Each process has threads, if it's a single threaded application it counts as one thread. Each thread is attached to a LWP or Light Weight Processor. The kernel then schedules the LWP time to run on the real CPU.

What's the end result of this? Solaris scales very well on boxes with tons of CPUs because everything is threaded, and some processes have tons.

Re:no surprise

[Anonymous Coward]

Did you really just say "Light Weight Processor" instead of "Light Weight Process" wrt LWP? So why should we believe you when you can't get your acronyms right, especially when the form you used doesn't make sense?

Re:no surprise

[Anonymous Coward]

Did you really just say "Light Weight Processor" instead of "Light Weight Process" wrt LWP? So why should we believe you when you can't get your acronyms right, especially when the form you used doesn't make sense.

He has the correct idea but not the correct concepts. GET HIM!

Re:no surprise

[Anonymous Coward]

There's a wicked job in Midtown I want bad.

Re:no surprise

[kerubi]

Correction: LWP in Solaris doesn't mean a "Light Weight Processor". It means quite exactly the same as a thread.

Re:no surprise

[Spl0it]

I thought solaris only supported 4 processors compared to linux's 16... Maybe I'm smoking something but wasn't that even mentioned in the article? 5 > 4 if not 6...let alone another 11 processors.

Re:no surprise

[Anonymous Coward]

Solaris supports at least 16 or 32 processer systems. Seem to recall seeing 64 or 128 too, but might be wrong on that.

Surprisingly good article

[sql*kitten]

This is a surprisingly good article for OSNews. Usually their reviews are limited to utterly trivial things like what the reviewer thought of the default colour scheme, or how easy it was to change the desktop wallpaper. But this one actually has some useful quantitative data in, and refers to things that workstation users actually care about, such as compile times. Whoever this chap is he should take over doing all the reviews from the girl (can't remember her name offhand) who usually does them, because she is pretty much clueless.

Re:Surprisingly good article

[vrt3]

Eugenia Loli-Query or somesuch.

Re:Surprisingly good article

[Anonymous Coward]

I agree with you that OSNews articles are usually pretty bad. But I don't agree that this one is very good. (Actually, I find it surprising that I even followed the link. Must be some failing in my nature.) There's a lot that's gonna be said about his "benchmark", and probably all of it justified. Let me just say that if I wanted to match Linux against any other operating system for any benchmark except ease of installation and trouble free operation I would not pick RedHat. Don't get me wrong, this is not a RedHat bash. I like RedHat and run it on machines I don't want to spend a lot of time administering and tweaking. But RedHat has been designed as a compromising operating system so that it will run on the broadest range of hardware and support the largest set of applications possible. It does a good job at those things; but at the cost of performance. Solaris, if designed to be an enterprise OS, does not have to support the broad range of hardware and software that RedHat does. For fair comparrison I'd have pitted Solaris against Debian or Slackware. But if it was my shop I'd run a customized and tested Gentoo implementation.

Re:Surprisingly good article

[Anonymous Coward]

Usually their reviews are limited to utterly trivial things like what the reviewer thought of the default colour scheme, or how easy it was to change the desktop wallpaper.

But Solaris has shipped with some of the ugliest-assed default color schemes in history. Ignoring this fact won't make it go away.

Re:Surprisingly good article

[Anonymous Coward]

I disagree. This article's benchmarks are very arguably skewed, and his rationale for optimizations is just awful. (Solaris sucks on x86 as much as linux 2.4 vs 2.6, so let's compare x86 Solaris with 2.4 linux??) Not to mention that Redhat's packages are still compiled for i386...

On top of that, the author's writing is all over the place, and not at all consistent. The part about "your momma [being] ugly" aside, he has also written a book for O'reilly, which received some very negative reviews. Terrible read, unfortunately. I'd avoid anything he writes, period.

Re:Surprisingly good article

[Anonymous Coward]

Solaris versus RedHat is a very fair comparison because RedHat's salesmen are out there every day selling against Sun in the "enterprise" space.

Realistically if your company is used to Sun's expensive support packages, you aren't going to switch to Gentoo.

Re:Surprisingly good article

[mbyte]

soso ... maybe the average level of OSNews is so low, BUT ... just look at their benchmarks and read carefully. They did benchmark gcc 2.95 against 3.3.1 ... everyone knows that 3.3 is slower (which might change in 3.4 )

Re:Surprisingly good article

[Haeleth]

> Just look at their benchmarks and read carefully. They did benchmark gcc 2.95 against 3.3.1.

Just look at their benchmarks and read carefully. What they benchmarked was gcc 3.3.1 compiled with gcc 3.2 against gcc 3.3.1 compiled with gcc 2.95. Not quite the same thing.

Re:Surprisingly good article

[Anonymous Coward]

I'd rather have a workstation that takes 30 minutes to compile instead of 5, it gives me time to have some coffee and a pretzel, and procrastinate about how slow is the hardware we have and how cheap is the company.

Re:Surprisingly good article

[nullity]

This is one of the least relevant and useful articles I have seen OSNews carry. So it would be nice to have my compiles done 1/6th faster... but not as important as my working environment. Which is the sort of thing Eugenia reviews.

People who are actually working and not just drooling over hardware tend not to care as much about small fluctuations in performance (servers are of course a different matter; though even there things like reliability are much more important). When you're working at a workstation, people can hardly tell the difference until you *double* performance.

"Quantitative" data doesn't necessarily mean "better" data if its not measuring something terribly important.

In any case, whether you think its useful to have tons of benchmarks are not, there are a million sites already doing them. OSNews is doing something different.

Re:Surprisingly good article

[Anonymous Coward]

he has also written a book for O'reilly, which received some very negative reviews.

Oh man those reviews made me laugh so hard I cried, especially the positive ones:

"Having been thrusted into a consulting position with a company who uses Alteon-based products, this book has been like a bible to me."

"thrusted"!

I feel sorry for that company.

Re:Surprisingly good article

[_|()|\|]

Redhat's packages are still compiled for i386

Last I checked, they're compiled with the flags -march=i386 and -mcpu=i686. I've yet to see compelling evidence that switching to -march=i686 would yield a substantial improvement.

Re:Surprisingly good article

[_|()|\|]

if it was my shop I'd run a customized and tested Gentoo implementation.

Not if you wanted support from Oracle, Sybase, IBM, or BEA. Red Hat Linux and, to a lesser extent, SuSE Linux, are the supported distributions, although Red Hat Enterprise Linux and SuSE Linux Enterprise Server are quickly replacing them.

Re:Surprisingly good article

[Anonymous Coward]

Good point for a real world shop. But please note I said "my shop". I deal with IBM, Oracle, BMC, etc everyday at my boss' shop. Their support ranges from poor (Candle) to mediocre (IBM, Oracle) to superlative (IBM,BMC). But in 15 years I can say they've only been indespensible once. Nine times out of ten I find what I want in groups.google.com. And in my shop I'd rely on quality systems programmers, quality application programmers, and open source software and NOT, I repeat NOT, support from any of those companies named. I've been doing big-biz computing for 15 years, Gnu at home since before linux, and linux since December of 91 and I can swear to you I wouldn't miss big business tech support for a second.

Big business tech support is hired for one reason only: So that middle level managers can explain to upper level managers that "Yes, there is a problem. No, I do not know what it is, nor how to fix it. But IBM has been contacted." It's a gawd awful expensive CYA pacifier.

Re:Surprisingly good article

[Anonymous Coward]

How very true.

But please note that I said "in my shop".

In my shop the programmers would only code in one language: the one they wanted. And they could buy any software they wanted to use: with their own money.

My shop probably wouldn't be profitable. But we'd have a hell of a time coding some great stuff. In my shop it would be shameful to say, "There's a problem and I don't know how to fix it but I've called XYZ Tech Support and opened up a problem ticket!"

Yeah, go ahead and mod me off topic. :-)

Re:Surprisingly good article

[Anonymous Coward]

funny, it looks ok on my green-screen Wyse-term...

um...

[GoNINzo]

Who bothers to use Solaris on x86? Oh yeah, idiots who don't understand the 'right tool for the job' philosophy.

Go with OpenBSD or Linux on x86. If you want to run Solaris correctly, get some ultrasparcs already. You always lose something when you skimp on your infrastructure to save a few bucks.

Re:um...

[Anonymous Coward]

exactly what part of going from zero ultrasparcs to n>0 ultrasparcs constitutes "a few bucks?"

Can I borrow a few bucks?

Re:um...

[n3rd]

Who bothers to use Solaris on x86? Oh yeah, idiots who don't understand the 'right tool for the job' philosophy.

But what if Solaris x86 is the right tool for the job?

It may not be often but it is at times.

Re:um...

[LurkerXXX]

Maybe you just understand what the job is.

Solaris for x86 has always been for training or utility. Cheap or free x86 versions mean people who want to eventually admin Solaris on Sparc equipment can easily get ahold of it to try without having the capital to purchase a Sparc just for learning purposes.

The other common job for it has been as a utility/admin machine. When you have 20 Solaris Sparc servers to admin, it's just easier and more consistant to also run Solaris on the admin's workstation. Why throw on another *nix with it's own individual setup and quirks when you can run something integrated and consistant. It doesn't really matter if the admin's box is 20% slower than an equivalent Linux machine, it still works fine to admin the big iron.

For some jobs, it is the right tool. Except to idiots who don't understand the job.

Re:um...

[ninejaguar]

According to this, you should go with Linux on x86.

= 9J =

Re:um...

[molarmass192]

I've played around with Solaris on x86. It's neat but if you think software for Linux is hard to come by, software for Solaris x86 is virtually non-existent. I'd be really really hard pressed to find a situation where I'd pick it over Linux for anything really. Come to think of it, it doesn't even make a good training tool for Solaris noobs since the hardware / BIOS is so different from the Sparc version of Solaris. Did I mention that it's neat?

Re:um...

[GoNINzo]

Have you maintained an x86/sparc environment? It's a pain in the ass. You have to do many things differently anyway, so why bother. Besides, the Blade 100's are only a thousand bucks so they are not that much to the 24k-48k you pay for the low end 280R's and such.

And if it's a utility machine, you can look at the netra x-1's, which are quite nice.

I've just seen too many people put up farms of x86 machines, brag about how they are running solaris, and spend all day trying to fight the fires with it. The TCO just doesn't make sense just so you can say 'We're running Solaris!'

It's more management's fault, not necessarily the people under them.

Re:um...

[LurkerXXX]

Yes, yes I have, but apparenty you didn't read my above post. Where did I suggest the right job for it was to run a farm of x86 servers? (Not to say it couldn't be, there are some folks with exceptional requirements who it might be the best fit for, but those are the minority).

I said it's for learning or as an admin's workstation. How many college students, or Windows/Linux admins are going to buy a blade just to learn the OS? Why do that when you can get a cheap/free x86 version? They can learn the OS, then only have to deal with the hardware specific details when you actually get an opportunity/cash to work on a real Sparc machine.

And as an admin workstation, you are don't just play with configurations on your own box all day. Maintaining a mixed environment when only your workstation and test machines are x86 is just not that difficult, sorry.

Yes, there are a few folks who run it just to brag about running Solaris, but that isn't the right job that I was talking about.

Re:um...

[haggar]

I disagree: Solaris x86 is an excellent NFS server. Rock solid, reliable, locks work as they f**** should and it's fast! Plus, it has one of the best storage management software bundled for free: Solstice DiskSuite, now re-labeled Sun Volume Manager. It has already cought up with Veritas on almost all features, plus it's easier to configure for maximum performance and rendundancy than Veritas (it is missing clustered diskgroups, but that shall come soon).

Re:um...

[Evil MarNuke]

You always lose something when you skimp on your infrastructure to save a few bucks.

Yep!! Like the support desk number! The power edge crashed. First time ever. Do you have Dell support number??? If it was the E-450 or E10-K, I would hit *2 on the speed dialer.

Re:um...

[Anonymous Coward]

if its posted on slashdot it MUST be true!

Modern system?

[brennz]

* from the article * Processor (2) Intel Pentium IIIs at 600 MHz, 256 KB cache Motherboard Intel L440GX+ RAM 512 MB PC133 ECC DISK (1) 9 GB Maxtor SCSI LVD SCSI Controller Adaptec AIC-7896 Dual Channel Video Cirrus Logic GD 5480 2 MB RAM *********** Was this test conducted in 1999 or 2003? ROFL. I think the hardware being tested should be modern to be representative. Throw some dual XEON systems or Athlon MPs in there. No legacy PLS!

Re:Modern system?

[grub]

Modern != common usage

I'd wager there are many more of these types of systems running important tasks than there are bleeding edge ones. The dot-bomb made people look at the bottom line ya know.

Re:Modern system?

[ebh]

There is actually a very good reason to benchmark on slow or deliberately pessimized hardware: The blink-of-an-eye test.

Many tests can be run 10,000 times to get good macroscopic benchmarks. But there are others (e.g., process startup) that can only be run singly, and if the amount of time (or other resource) it takes to do it is below the resolution of whatever you're using to measure it, you end up with meaningless results:

$ time true
real 0m0.00s
user 0m0.00s
sys 0m0.00s

On slow hardware there are fewer cases of this, so although you could design more elaborate tests to avoid this problem, old slow hardware is a cheap effective workaround.

Re:Modern system?

[cyb97]

While I agree to some point, there are issues of running tests on old hardware.
CPUs change, a Pentium4 (or even a Xeon) have more ops than the P3 does, so a proper compiler would be able utilise the processor differently, hence ending up with a test that compares apples and oranges instead of two old apples...

BSD?

[millahtime]

Does anyone know how any of the BSD OSes stack up against this?

Re:BSD?

[Anonymous Coward]

Dude, BSD doesn't even run on multiple
processors on x86. Why waste your time?

Re:BSD?

[Anonymous Coward]

Wrong

Re:BSD?

[millahtime]

actually freebsd does run on multiple processors. You can find more info at http://people.freebsd.org/~fsmp/SMP/SMP.html

Re:BSD?

[ninejaguar]

There is this recent benchmark of BSD vs Linux. You can also find it here.

= 9J =

Re:BSD?

[Anonymous Coward]

It stacks up 6 feet under, because BSD is dying.

Totally Serious Post

[Anonymous Coward]

When will the weekly Sun bashing on Slashdot end?

Re:Totally Serious Post

[brennz]

I don't think this is Sun bashing.

It is widely known that Sun has waffled on x86. Now because their market share is being eaten up, they are reconsidering years of mistakes...I was one of those admins that had to struggle with Solaris x86.... I woke up though, and went with GNU/Linux + *BSD.

Can Solaris compete with Linux/*bsd on x86? Try a modern distro, and form your own opinions...

Solaris scales better and has more features.

[Anonymous Coward]

Solaris users, sysadmins, and developers never cease to tell me how great Solaris is, for its scalability and features that Linux has yet to catch up on.

Sun isn't the leader in the Unix industry for no reason.

Re:Solaris scales better and has more features.

[shaitand]

"Solaris users, sysadmins, and developers never cease to tell me how great Solaris is, for its scalability and features that Linux has yet to catch up on."

Linux users, sysadmins, and developers never cease to tell me how great linux is, for it's scalability, features, applications, free availbility of sourcecode and speed that Solaris has yet to catch up on.

"Sun isn't the leader in the Unix industry for no reason."

Linux isn't the overall leader when all computing markets are added up for no reason.

Gee this is a real revelation, we've determined that those who like solaris say it rules, and those who like linux say it rules ;)

Smart Move...

[ShieldW0lf]

Put up a review stating that Suns OS sucks, and cap it off with this statement:

"In fact, it's possible installing Solaris x86 on my dual-processor box, even if I disabled one of the processors, violates the evaluation license that Sun offers Solaris x86. Oops."

Smart move, dumbass...

It's a sad state of affairs...

[Anonymous Coward]

... when this is the best article OSNews has had in a long time.

Tony makes the it's/its mistake a half dozen times. Nice editorial work, OSNews!

Your well on you're way to having a professional website!

Re:It's a sad state of affairs...

[Anonymous Coward]

Were you trying to be funny?

Re:It's a sad state of affairs...

[Anonymous Coward]

They also did the "affect/effect" thing. Damn that makes my brain hurt.

Solaris: Time machine to the 1980s

[GGardner]

Running Solaris on X86 is like going back in time to the 1980s with all the ancient Unix utilities -- it's got the ancient vi, not VIM, which is annoying when you need things like multiple undo or multiple windows. The awk/nawk are ancient, and it doesn't ship with perl (last time I checked). Ditto for most of the Unix shell programs.

The first thing I do when I get a Solaris system is to install a whole heap of GNU utilities, all of which come with any of the Linux distribution.

bah

[DNS-and-BIND]

Who the heck runs Solaris on a crappy dual-processor machine? Solaris doesn't really even begin to show benefits until it gets at least 8 processors. It just keeps going up, and up, and up.

Solaris/Intel is just a toy that grabs a few extra customers that Sun would have lost otherwise. Boy, you should see it when linux noobs get their hands on it. They get really angry when you tell them "your hardware must be listed on the Hardware Compatibility List". I've seen venomous diatribes directed at "sucky" Sun and its "sucky" OS for not having video drivers for whatever the most expensive game-playing graphics board is these days. And if they actually get the system to install and they see CDE...oh man.

I don't hang on #solaris any more, but damn we would get the same reactions over, and over, and over about Solaris/Intel.

Re:bah

[Anonymous Coward]

I must be the only person who actually *likes* CDE. Granted, I've not done too much trying to customize it or use it as a home desktop, but for work/development, it's actually rather nice and simple.

Re:bah

[Lumpy]

Who the heck runs Solaris on a crappy dual-processor machine?

Ahh the rant of a non-knowing person...

Thousands, if not tens of thousands use Solaris on dual and ...GASP! Single processor systems!

In fact the company here has 12 sun workstations running solaris as data collection servers that are going to be replaced here in 180 days. (giving me a few Sun UltraSparc toys to play with as I see fit later)

Solaris is used mostly for single and dual processor systems. there are significantly less 8 processor and up solaris systems in existance.

Re:bah

[n3rd]

I've seen venomous diatribes directed at "sucky" Sun and its "sucky" OS for not having video drivers for whatever the most expensive game-playing graphics board is these days. And if they actually get the system to install and they see CDE...oh man.

So true. Solaris is built for work, not for play.

Re:bah

[Zan Zu from Eridu]

Solaris/Intel is just a toy that grabs a few extra customers that Sun would have lost otherwise. Boy, you should see it when linux noobs get their hands on it. They get really angry when you tell them "your hardware must be listed on the Hardware Compatibility List".

So basically, what you're saying is: Sun is trying to grab a few extra customers, but doesn't support them very well (no drivers for boxes you can actually buy in stores today, very late security patches, crappy toolset, the list goes on), so anyone foolish enough to buy a Solaris licence that previously ran a free OS with better support shouldn't whine or get angry if he gets what he deserves.

Solaris on Intel should convince people to make a switch to "real" Solaris on "proper" hardware, because a lot of people are going to try and test Solaris on boxes they already own (meaning mostly Intel boxes). If you piss of a real customer long enough by telling him he's not considered important enough to be offered support, you'll make an enemy for life; he's never going to buy Sun products again, no matter good they may be.

Re:bah

[iggymanz]

funny in the pilot Linux/Oracle project I just did for a *very large* midwestern city that a 4 way Intel Xeon box outperformed a 14-way SunFire 6800 by a factor of 3, even with the much smaller 2G SGA & contortions an Intel chip has to go through to get to >4GB memory chunks

*something* is killing Sun, and they can't survive on just their 8-way or more system sales. Wait until the finer grain SMP in Linux & FreeBSD get perfected.....

Re:bah

[Anonymous Coward]

Let me guess...the linux admins were adminning the sun box...

Re:bah

[DNS-and-BIND]

Non-knowing? If your shop runs Solaris, and doesn't use the bigger boxes, you're wasting money. Just buying a bunch of single-processor workstations is a complete misappropriation of funds. Lesser machines can do the job quite well at a fraction of the cost. The whole reason for the UltraSPARC line in the first place was to give developers something to work on that runs just like the real machines.

Re:bah

[Lumpy]

The whole reason for the UltraSPARC line in the first place was to give developers something to work on that runs just like the real machines.

Yup, and it's the same mentalility that makes us a MS shop now. (another mis-appropriation of funds)

they were purchased 4 years ago for developers, that project was cancelled and we used the hardware for some other use.

MS couldn't do the job (and still cant) and the new linux machines going online in their place will do the job... much to the displeasure of the MS loving MIS director.

anyways, they make many MANY more ultraSPARc stations than servers with 8 processors or more, so the statement is still Wacked.

Re:bah

[iggymanz]

no

Re:bah

[_damnit_]

Well, I guess it really depends on what you were doing with the boxes and how you were benchmarking. I have seen exactly the opposite in a recent case. A customer brought in some Dell's that were supposed to do the work of two Sparcs of equiv specs (dual procs both). The customer (a HUGE Intel backer) was stunned to find out one sparc box did the work of more than two x86 boxes. YMMV. It really depends on what you are doing with your boxes. That is why it is really difficult to generalize about performance.

Re:bah

[Brandybuck]

Solaris on Intel should convince people to make a switch to "real" Solaris on "proper" hardware

They are, and always have been. As one example, currently you can download Solaris/Sparc for free, but to get Solaris/x86 you have to pay $99.

If Apple ever seriously considers releasing an OSX/x86, they need to take a long hard look at Sun. By releasing Solaris/x86 they instantly commoditized their software while changing their hardware's perception to "expensive". Not a good thing.

Re:bah

[Anonymous Coward]

-1, Flamebait

Re:bah

[iggymanz]

very true. In this case, there were some industry standard benchmarks run, and then some of the client's processes. The main point is that 32 bit Intel boxes are a big threat to Sun's low and mid-range for many applications. As Linux & the open source BSD's gain High Availability, distributed lock manager, finer grained SMP (all probably in less than 2 years), I think the commercial Unixes will be killed off.

Re:bah

[as_dup]

People are always trying to compare linux/intel against sparc/solaris or intel/solaris on 1 and 2 processor boxes and it simply is not fair. When you consider what an OS has to do to keep scaling as you add very large amounts of memory and more processors, add separate run queues (one for each processor), fine grained locking primitives and lots of other overhead to keep scaling in a near-linear fashion, that on a single processor box your gonna get your goose cooked when compared to a less scaleable, more simplified OS (which is exaclty what linux is ..). Please don't drag an 18 wheeler out on the freeway and race it with your porsche unless your gonna strap on a trailer with 50,000lbs of carrots in the back to the porsche! The Enterprisability of Linux for vertical scaling applications still isn't there. 1) There is no path_to_inst like functionality to make devices stay put. If you add/delete a card, your devices move around. This just doesn't cut it from the managability aspect. 2) Unless you find that specific distro version from that specific vendor, and strip a ton of stuff off, your not going to get as stable as an OS as Solaris, period. Save the rants about your 300+ days of uptime because just because you can make your linux distro stable, doesn't mean that linux *is* as stable as Solaris, HPUX or AIX.

Re:bah

[_damnit_]

For selfish reasons, I hope you are wrong. If you are, does it make you feel any better that RedHat will be your system integrator? Someone has to set a standard that everyone can write for. Oracle, PeopleSoft etc. write their code to RHAS which has certain libraries, java, and other pkg dependancies. It seems to me that everyone is trading one landlord for another. I would feel immensely better if a standards org with real teeth and community backing would emerge and decide what milestones everyone will target. That way SUSE, RH and even IBM could release a Linux distro that had the same kernel, glib, gtk+ etc as everyone else.

In any case, you are missing the direction most big Unix vendors are going. Linux does not have partitioning. I haven't heard any plans for that from anyone (I could be wrong though). How about proc affinity or processor sets? IBM and Sun seem to be heading toward what Sun calls N1. Moving load to any machine in your datacenter as needed. It's essentially virtualizing your systems to enable seemless, automated maintenance of all your hosts. I know both are going that direction. I know some customers who are foaming at the mouth for this. You need an integrator for this. Whether it's IBM, HP or Sun, who knows. It won't be Dell (no R&D) and it won't be MS (it would only work on MS products).

So there are reasons Sun et al aren't going to be lapped by Linux. Linux is great for small servers. Linux is great if you have a rack of servers. Linux is not robust in the manner I mentioned earlier on large systems. The larger the system, the more often something will break that requires replacing hardware. Try telling a CIO that the ERP cluster has to come down so you can change a DIMM. I'd rather DR that board out and replace it w/o the downtime or talking to a very unhappy CIO.

I digress....

Re:bah

[iggymanz]

yes, standardization of Linux as you speak of would be a fantastic thing

Oracle can already run on other flavors of Linux than Red Hat, and at least one of the open source BSD's. As for SUPPORTED distros, there's 3 others besides RedHat - Suse ES, Conectiva ES, TurboLinux ES.

Robustness is here for the big enterprise app with Oracle RAC. With 10g we'll have Oracle grid computing for Linux: "unbreakable Linux" is what Oracle calls it.

Processor affinity is here already in 2.6, and also exists in Red Hat advanced server

Talking about N1 type computing is interesting, but it isn't here yet. Who's to say Linux won't evolve a similar thing FASTER than the commercial vendors can??

There are already boxes that are sold with hot swap memory, network card failover, SMP CPU failover, etc. for Linux.

I'm telling ya, Linux is rocketing right on by the commercial Unixes.

Not really a smack-down

[Anonymous Coward]

The article claim sun said " better, safer, and more stable alternative". Then it just goes on to do a straight line speed comparison. What would have been more intresting if they wanted to dispute that claim was to put it under some really serious loads to see which responded better etc. Which OS keeps serving web pages with a load of 12?

Re:Not really a smack-down

[Bull999999]

Maybe you should checkout the Slashdot FAQ and see for yourself what kind of OS Slashdot runs.

You are missing the point

[ultrabot]

Linux has no place on Server. Period. Solaris is industrial strength, mature, stable and rock-solid. It has the "secret sauce" that even SCO doesn't know of. It brings Value to Enterprise, unlike that hobbyist jalopy OS. After all, it's about Systems. Now you can use commodity x86 hardware with the OS made in heaven.

And yeah, almost forgot, lots of mobile phones run Java.

Re:You are missing the point

[mydigitalself]

"secret sauce"
well substantiated, keep it up.

Re:You are missing the point

[Anonymous Coward]

Right ... A good troll should outragous, but still subtle enough so as to seem genuine. Keep trying ... you'll get it right one of these days.

Re:You are missing the point

[Anonymous Coward]

To the dildos who replied to this as a troll, I think it was meant as a joke. Mod as "funny"

Re:You are missing the point

[Anonymous Coward]

Are they letting any idiot be a moderator these days? It's a joke people, you know: Ha, ha, funny!

Re:You are missing the point

[Anonymous Coward]

Oh, SO very funny..

NOT!

Goodness, with "humor" like this, perhaps a tag would be appropriate. At least then I'd know I was supposed to laugh.

Sheesh

Re:You are missing the point

[Anonymous Coward]

You modded this insightful? It's nothing more than blather. Since when did name calling and unsupported assertions become insightful? Whoever mods these things are idiots. I now wear my constant "0: It's beneath your dignity to read this" mod as a badge of honor.

I've used both Solaris and Linux, and, in general, I'd rather use Linux even on SPARC systems. The system functions aren't strewn over a half-dozen libraries, the utilities actually do what they do on most Unix systems, and I don't have to spend lots of time figuring out where Sun hid the system configuration data with each new release.

Sorry to be making specific criticisms. You guys just lie down for a while until the dizziness passes.

Re:You are missing the point

[dcavanaugh]

"Linux has no place on Server. Period."

Everywhere you see low-end servers (especially Windows) -- those are opportunities for Linux. Hell, even the desktop is up for grabs.

Re:You are missing the point

[ultrabot]

well substantiated, keep it up.

Come on, people, it was *obviously* a joke. Or a parody, at least.

Mod this down please

[Captain Kirk]

Very unfair point. The girl at OSNews produces 3 to 4 stories about operating systems a day. Of course a lot of them are trivial - lets face facts - big news in the OS world happens once every year or so. Are we really saying she shoudl only update the site twice yearly?

Re:Mod this down please

[Anonymous Coward]

Are you saying that if I talk a lot, it does not matter how poorly I express myself?

Eugenia's articles are painful to read.

The old "no Linux strategy" quote

[ChrisRijk]

Appears in this article again.

I think what he meant was that "we don't have a Linux strategy" means "we don't do Linux only products or development" and "Linux doesn't play a role on the server" means "Don't write/develop to the OS" - ie could say the same thing about Solaris if you contort it that much. Possibly being overly generous, but it's about the only logical thing I can think of. (unless you assume the guy's gone nuts.)

Last month Jonathan Schwartz did a fairly in depth response on his thoughts on Linux on general, though nobody seems to have reported it:
Schwartz Seeks to Clarify Sun's Linux Strategy

He doesn't really clarify his statements, but kinda starts from the beginning. Among other things he talks about ISV support, open standards, Debian (which he's a "big fan" of), and also indemnification - which Sun offers both on Solaris and Linux.

Re:Solaris: Time machine to the 1980s

[brennz]

great points! It almost makes you think Solaris is behind the times......

How to look scalable...

[Anonymous Coward]

There's an old trick to appearing scalable: Make your single-processor performance suck. Then almost any parallelization overhead will disappear, and you'll "scale" linearly.

Many papers make this mistake. If you ever see scalability comparisons without pure time comparisons, don't trust the results.

Re:How to look scalable...

[Anonymous Coward]

There's an old trick to appearing scalable: Make your single-processor performance suck.

How about this idea: Sun forks Solaris x86 to a single processor optimized version and a SMP optimized version?

Re:How to look scalable...

[4of12]

an old trick to appearing scalable:

Trick #2: Pick an your application so it is embarrassingly parallel.

Re:How to look scalable...

[shaitand]

start multiplying by 2 (solaris base speed), now start multiplying by 3 (linux base speed). It won't take long before you see it would take a pretty serious scaling loss to compensate for the fact that you start with a lower base number.

2 and 3 of course are meaningless numbers pulled out of my arse, just something to plug in and see what kind of difference it can make being faster per processor even if you lose slightly more of your base speed for each processor added.

On Linux

[bogie]

"Also, let me really clear about our Linux strategy. We don't have one. We don't at all. We do not believe that Linux plays a role on the server. Period. If you want to buy it, we will sell it to you, but we believe that Solaris is a better alternative, that is safer, more robust, higher quality and dramatically less expensive in purchase price"

Wow. What a ringing endorsement. Tell me why anyone would buy Linux from people like that when you can deal with IBM.

Salesman rule #1 - Never Ever talk bad about any of the products you sell even if you think they are crap. I'm sure the guys in the Linux marketing and sales department at Sun are oh so happy about that quote.

Re:Solaris: Time machine to the 1980s

[mark_lybarger]

the same is true afaict for solaris on sun hardware. the utilities available on solaris are laughable. maybe they've updated them on solaris 8, or 9.

Re:Solaris: Time machine to the 1980s

[Anonymous Coward]

I've got news for you... that's because Solaris/SunOS is UNIX, not Linux. UNIX includes vi... Linux doesn't. Liunx comes with a clone, vim. vi will run in a tiny amount of space. vim won't. Most old-skool sysadmins expect old-skool vi, not vim. They are called "standard". Believe it or not, the GNU people pull a Microsoft with this and add all their little "extensions" to it. This is nonstandard.

Another thing that pisses me off is that /bin/sh is really a symlink to /bin/bash on Linux. This, again, proves that Linux isn't a real Unix. This is nonstandard, and it breaks Bourne shell scripts.

We switched over to Linux PCs from Sun Boxes

[Anonymous Coward]

Solaris was awful, it was too ancient and didn't even come with Perl. It was like using some 1970s system. When we upgraded to Linux we got rid of our Sun systems (everything from an old sparcstation 10 to a few Enterprise 4500s) by taking them out back and bashing them with hammers and baseball bats. It was hilarious.Solaris was awful, it was too ancient. It was like using some 1970s system. When we upgraded to Linux we got rid of our Sun systems (everything from an old sparcstation 10 to a few Enterprise 4500s) by taking them out back and bashing them with hammers and baseball bats. It was hilarious.

Whereas Linux is still in the 1980's

[Anonymous Coward]

command line ? shells ? compilers ? terminals ?

sounds like its still stuck there, even my Atari had a GUI

Re:Whereas Linux is still in the 1980's

[N1KO]

How did you manage to find a system that doesn't need compilers? Do the programmers for your operating system do everything in hex editors?

Re:Whereas Linux is still in the 1980's

[Spl0it]

Who is the bright guy that mode'd up the Parent.. can we say Troll?

compare many users

[Anonymous Coward]

my longterm experience: many users on a slow solaris sparc is much faster than many users on a fast linux x86 and one user on linux x86 blows away one user on sparc solaris.

so what is it? x86 v. sparc or linux v solaris? they dont answer this question vis a vis multi user configs.

Re:Solaris: Time machine to the 1980s

[Anonymous Coward]

i don't see why solaris is suddenly 'out of the 80s' because you can't use pico or vim. maybe you should explain this. not to start an editor war but vi is for simple tasks while you should use emacs for complicated edits/development. In general running GNU stuff on a sparc box is like taking a white dress and tye dying it. this article is kinda useless. i would like to see the performance of the solaris compiler in solaris and gcc in linux. the binaries from the solaris compiler ALWAYS run faster. solaris is a better OS if you're serious about real development or want a really rock-solid platform. yes it takes more work than installing a redhat package but i think its more stable in the end just my opinion, but linux is fine for home users.

not newsworthy - move along

[Anonymous Coward]

this is boring bs pretending to be /. worthy.

Re:For ramblings on "Oracle on Solaris or Linux?".

[GoofyBoy]

Interesting, but I don't most of the points apply anymore.

What we have here ...

[Anonymous Coward]

... is a comparison of the 'lesser of two evils'

Personally, Linux is pisswater compared to Solaris.

Netfilter

[Ianoo]

I don't know why this guy spends so long complaining about Netfilter. If he wants ease-of-configuration, then download something like Shorewall. I am not a Linux newbie, but I am fairly new to software firewalls. However when I moved one of my boxes out from behind its hardware firewall/router for a few days, I downloaded Shorewall and had it up and running in less than 10 minutes, then it took me about a minute to work out how to open port 22 for SSH.

This Guy Is An Idiot

[fupeg]

This guy states that Linux is the clear winner in the enterprise software space, but he states this based on a comparison that uses Red Hat 9? If you want to run enterprise software on Linux, such as Oracle or BEA, you can't run Red Hat 9! You have to run Red Hat AS. Oh you can get Oracle to run on RH9, but you will not get any support from Oracle if you do it. The fact that the author left this little fact out indicates that either he doesn't know this or that he left it out intentionally. So either he is an idiot or biased. I will go with idiot, though the article would support either conclusion (or both!)

Re:This Guy Is An Idiot

[commander salamander]

What are you talking about? The benchmarks he ran reflect the linux kernel, fs, and glibc performance. He says that right in the section about choosing a linux distro.

Of course you will shell out the $$ for RHAS if you are running Oracle, but that only gets you support. It will benchmark the damn same. Since you missed that little fact, you must be an idiot or biased. I will go with idiot, though your post would support either conclusion (or both!)

Re:This Guy Is An Idiot

[fupeg]

The article was more than a collection of benchmarks. The author spent a great deal of time talking about things like installation, drivers, desktop features, and available software. My comment was in response to the author's remarks about the availability of commercial software for Solaris x86 vs. Linux. His remarks showed a lot of ignorance by talking about enterprise software like Oracle being available for Linux while the distro he chose for his comparisons was Red Hat 9. The whole point is that Oracle is NOT available for Red Hat 9 only for Red Hat AS.

And you showed your own ignorance by stating that:

It will benchmark the damn same.

Red Hat AS's kernel is the same as RH 7.2 plus some other bits that Red Hat as added. It is far shy of the 2.4.20+ kernel being used in the benchmarks, which implements full POSIX threading (and contains some of those naughty bits for SMP systems, like the ones in the test, that has SCO suing everybody and their brother) among many other things. Let's think about a place where an improved thread model might make a big difference ... hmm, maybe in a web server load test? That might involve a lot of threads, what do you think? I would guess that RH 9 would offer significant performance benefits over RH-AS in such a test, but maybe not. However, it would be truly ignorant to guess that they would benchmark the same.

Re:This Guy Is An Idiot

[platypus]

ty bits for SMP systems, like the ones in the test, that has SCO suing everybody and their brother) among many other things. Let's think about a place where an improved thread model might make a big difference ... hmm, maybe in a web server load test? That might involve a lot of threads, what do you think?

Sorry, but you're talking out of your ass. He tested Apache 1.3.28, and this apache doesn't use threads at all. Go read up about it. This is the reason why apache was aways so much faster on linux than on solaris or windows (talking UP machines here), because linux' processes are cheaper. And that is the reason why the apache people heavily changed their code to allow for threading. Again, inform yourself about apache 2.

Btw. ironically the reason mainly told for making this change to apache was windows performance, seems solaris isn't important enough (ok, maybe on solaris a commercial http server like netscape is prevalent anyway),

Re:This Guy Is An Idiot

[fupeg]

You're right about Apache. My mistake for not realizing they were using a 1.3.x version of it. I cannot comment on the reasoning for making Apache multithreaded, besides just being better from a software architecture standpoint, but it is a moot point. I think you just provided the reason why Apache performed so much better on Linux than on Solaris x86, a pity the author of the article wasn't as knowledgeable as you on the subject. Of course that all goes back to my original point about the author's intelligence, or lack thereof. Your comments do little to discredit my claim that benchmarks done against RH 9 would not be identical to benchmarks done against RH-AS, though given your insights, I would agree that RH-AS would still outperform Solaris x86 when it comes to load testing against Apache 1.3.x.

Red-headed Solaris

[SARSpatient]

Neither Sun nor Sun resellers really push Solaris x86. This includes during their presentations to customers interested in high performance x86 clusters solutions. Yes, the alternative is always there, showing the Solaris logo along with a RedHat one under available operating systems. Even with clients where Solaris x86 might make sense, Sun salespeople skirt around the issue of O/S and never press their own version. Aside from Sun support (which IMO is really good), would there be any benefit to switching to Solaris? Everyone knows that it's not a core product or moneymaker for Sun, even their own sales associates who definitely know which products to push, and which to let slide. For now, I would leave Solaris x86 as a novelty, at least until Sun itself proves it has feature enhancements outperforming a Linux installation (especially on their own hardware).

Re:Red-headed Solaris

[sudohnim]

Aside from Sun support (which IMO is really good), would there be any benefit to switching to Solaris?

How about this:

Client wants to build an application of some sort. The resource requirements for the app will start small and may stay small, but resource demand from usage of the app may also grow quickly. Consultant recommends writing the app for Solaris x86 as this will allow for inexpensive hardware and OS aquisition fees. If the app grows, it can simply be recompiled for Solaris SPARC where its resource consumption can scale linearly with hardware additions.

This seems much less expensive compared to writting the app on Linux x86 and then having to port/rewrite it for Solaris SPARC (or AIX or HP-UX, etc).

Sun is indifferent to the x86 Solaris.

[reporter]

Sun Microsystems is indifferent to the x86 Solaris. Sun just posted a loss of $290 million for the last quarter. On an annual basis, the loss amounts to $1.2 billion. The managers at Sun have seen the writing on the wall: the future is Linux.

It is unlikely that Sun will do anything to optimize Solaris for x86 here in the USA. There may be some optimization work at Sun's R&D center in India, but basically in the USA, Sun is conceding to Linux. Linux is backed by IBM, and IBM and Linus are cooperating to make Linux a rock-solid product that meets 6-sigma reliability. Right now, Linus is concentrating on making Linux as stable as possible instead of adding more widgets and gadgets.

The penguin shall rule the world!

... from the desk of the reporter

Re:Sun is indifferent to the x86 Solaris.

[Emil Brink]

Oh my $DEITY! You just took a number, like 290 million, and multiplied it by what seems to be the number four! This is unprecedented, what do you have over there to help you, a computer?! I'm shocked. Shocked, I tell you.

Re:Sun is indifferent to the x86 Solaris.

[n3rd]

Sun Microsystems is indifferent to the x86 Solaris.

As you can see from the PR, and even the stories on Slashdot they seem serious about it again.

Sun just posted a loss of $290 million for the last quarter. On an annual basis, the loss amounts to $1.2 billion.

We're all familiar with basic math. So that means with their 5.5 billion in the bank they can last another what, 4 years? With articles starting to pop up about increased IT spending in 2004 who knows. Oh, and they still had revenue of 11 billion in FY03.

The managers at Sun have seen the writing on the wall: the future is Linux.

Where did this come from? Source or more information please.

There may be some optimization work at Sun's R&D center in India, but basically in the USA, Sun is conceding to Linux.

Great India bashing, looks like along with Sun bashing you've got a history.

Please elaborate on the "Sun is conceding to Linux" statement. What do you have to back this up?

Linux is backed by IBM, and IBM and Linus are cooperating to make Linux a rock-solid product that meets 6-sigma reliability.

Woah, 6 Stigma? Do you have more information about this or is it just a figure of speech?

Right now, Linus is concentrating on making Linux as stable as possible instead of adding more widgets and gadgets.

Yeah, working on getting the final 2.6 out will do that to a person. I think every Slashdotter knows that's what he's up to.

Re:Sun is indifferent to the x86 Solaris.

[jedidiah]

BAH!

If Sun were really serious about x86 they would have an 8-way x86 box for sale at store.sun.com today.

What this "benchmark" really points out is what sort of pathetic x86 Sun is trying to sell. WHO CARES about an overgrown desktop.

If you aren't benchmarking Linux against Solaris with at least 4 cpus you're simply wasting time.

Re:Solaris: Time machine to the 1980s

[Alrescha]

"The first thing I do when I get a Solaris system is to install a whole heap of GNU utilities, all of which come with any of the Linux distribution."

They come with the Solaris distribution as well. Not Sun's fault if you don't install them.

A.

Re:For ramblings on "Oracle on Solaris or Linux?".

[Anonymous Coward]

Especially because Oracle for Solaris x86 doesn't exist anymore (according to OSNews).

Re:Solaris: Time machine to the 1980s

[n3rd]

it's got the ancient vi

Solaris works for consistency and having a plane jane vi might be a good thing even if vim is better.

The awk/nawk are ancient

I don't use awk often, what's up with them?

and it doesn't ship with perl (last time I checked). Ditto for most of the Unix shell programs.

Yeah it ships with perl since Solaris 8, same with bash, tcsh and zsh.

The first thing I do when I get a Solaris system is to install a whole heap of GNU utilities, all of which come with any of the Linux distribution.

Yep, they're nice to have and don't take up much space if you pick and choose correctly. Here is everything you'll need as far as OSS utilities on SPARC/Solaris are concerned. They can be downloaded as a CD image or individually and are in Sun's package format.

Hi got to be kidding

[iamacat]

Processor (2) Intel Pentium IIIs at 600 MHz, 256 KB cache
Motherboard Intel L440GX+
RAM 512 MB PC133 ECC
DISK (1) 9 GB Maxtor SCSI

I bet Solaris is designed to run on more serious hardware. I bet DOS apps will run even faster than Linux on this box, even without taking advantage of dual processor.

Re:Hi got to be kidding

[pmz]

I bet Solaris is designed to run on more serious hardware.

I'd consider a dual Pentium III with a SCSI disk a dandy candidate for Solaris if there's no funds for getting a v210 or v60x.

Also, don't forget that Solaris/SPARC and Solaris/x86 are nearly the same code base (differing on drivers and assembler stuff, of course), meaning that Solaris/x86 fits very well into a Solaris/SPARC infrastructure.

Re:Hi got to be kidding

[MKalus]

National healthcare will be run with the fairness of the IRS and the efficiency of the DMV.

Actually there have been studies done that showed that public healthcare is more efficent than the private sector.

Go figure.

Re:Hi got to be kidding

[pmz]

Those studies ignore that nationalized heathcare is another stepping stone towards tyranny, which is more expensive than any possible gain of efficiency of a public plan. When the federal government has a centralized database of the health history of every citizen, just wait and see the legislation spew forth managing even more aspects of our lives. Also, watch as people who refuse to pay the added payroll taxes get sent to prison for their insolence.

In a free country, the government has no role in directing the individual health choices of its citizens. Imagine the irony of politicians spouting universal health care while also spouting that the USA is the land of the free. It's enough to make the founders of the USA to spin in their graves.

Power begets tyranny. It's sad how easily people forget that.

Re:Hi got to be kidding

[Brendan Byrd]

I guess Canada is just a suffering tyranny of a country. The real tyranny is mega-corporations in charge of a billion dollar industry, and like it or not, the system has a "role in directing the individual health choices of its citizens". Unforunately, since it's a system corrupted by money, the individuals are the ones that suffer.

The current state of capitalism is enough to "make the founders of the USA to spin in their graves". If our founding fathers saw the sorry state of capitalism today, they would have pumped Alexander Hamilton and Adam Smith full of bullets with their old-style gunpowdered pistols for even suggesting the idea. (Right now, I fear for the Russians 25-50 years down the line, when they learn how to become as corrupt as we are.)

Re:Hi got to be kidding

[pmz]

The current state of capitalism is enough to "make the founders of the USA to spin in their graves".

Why? The USA is still the most prosperous and inventive country on the planet, largely due to the delay between invention and regulation. When the regulation occurs, so does stagnation.

Corruption is temporary as it is unsustainable in a free market, because it has no financial basis (only so much money can be extorted from a community or market before it collapses and everyone flees). Finanical justice is justice nontheless. The government also has a nominal policing role to ferret out what corruption it can, but, ultimately, the government cannot catch every bad guy, no matter how hard they try.

Re:Hi got to be kidding

[Brendan Byrd]

Why? The USA is still the most prosperous and inventive country on the planet, largely due to the delay between invention and regulation. When the regulation occurs, so does stagnation.

Invention to benefit oneself only applies to oneself. Invention to benefit the community applies to the community. When a person invents only to get money, s/he will only works as far as to get that money and work no further. A person who is committed to actually serving the community will work as hard as s/he can.

The government also has a nominal policing role to ferret out what corruption it can, but, ultimately, the government cannot catch every bad guy, no matter how hard they try.

The government cannot catch every bad guy because they are one of the bad guys themselves. They are being paid by the corporations they try to "catch", and happily pass laws that the corporations require (whether it damages the public or not). (Examples: MS's paultry anti-trust settlement, DCMA, etc., etc., etc.) When the government is immune to corruption, the system will work at its peak.

Re:Hi got to be kidding

[Mr. Piddle]

When a person invents only to get money, s/he will only works as far as to get that money and work no further. A person who is committed to actually serving the community will work as hard as s/he can.

Most people are motivated by money. Few people are motivated out of love for community. One of the exceptions to this is doing things out of love for family, but families typically have only a few people in them. Most people would rather not take on the burdens and responsibilities that others can't outside of the family unit...unless, of course, there's money in it for them.

Selfishness is a fact of life, and, as long as our economic and legal systems take this fully into account, we'll be just fine. Communistic and socialistic idealism is founded on an impractical and naive hope (i.e., easier said than done).

Re:Hi got to be kidding

[MKalus]

When the federal government has a centralized database of the health history of every citizen, just wait and see the legislation spew forth managing even more aspects of our lives.

Having grown up in Germany I can't really say that the Government has a centralized DB. Actually it goes a lot further than this, documents have to be destroyed after 5 years and there are bodies who are controlling this, so from that stand point my privacy is better covered in Europe than it is in the US.

Also, watch as people who refuse to pay the added payroll taxes get sent to prison for their insolence.


They usually don't send you to prison for not paying your taxes (besides, when you're employed your employer is automatically taking this from your paycheck, if you're self employed you can "opt-out" since a couple of years and privately insure yourself, usually it is more expensive though.

In a free country, the government has no role in directing the individual health choices of its citizens.

Believe it or not, but the Government never told me who I can see and can't see. While I lived in the US though this was a completly different story, I had to check with my HMO before I could see a doctor I wanted to see.

Amazing, isn't it? I had so much "free choice" if I had a couple of thousand bucks.

Imagine the irony of politicians spouting universal health care while also spouting that the USA is the land of the free.

I don't think you've ever experienced that system, have you?

As someone told me the other day:

If a politician wants to win an election in the US, he speaks out against universal healthcare, if a politician wants to win office in Canada he better have a good plan for health care.

It's enough to make the founders of the USA to spin in their graves.


I am sure that if you connect a generator to theirt graves right now you could power half the country.... If they could see what their ancestors have made out of their dream they would rotate so fast you wouldn't believe it.

Power begets tyranny. It's sad how easily people forget that.

In what way does that relate to healthcare?

M.

Re:Hi got to be kidding

[MKalus]

Selfishness is a fact of life, and, as long as our economic and legal systems take this fully into account, we'll be just fine. Communistic and socialistic idealism is founded on an impractical and naive hope (i.e., easier said than done).

Well in that case: Why not give everyone a knife, take away with all those laws that try to prevent that we kill each other and let the strongest one win?

Would solve a lot of problems, no?

Re:Hi got to be kidding

[pmz]

Why not give everyone a knife, take away with all those laws that try to prevent that we kill each other and let the strongest one win?

Repealing the anti-murder laws is arguably a violation of the First Amendment. Neither extreme of anarchy or socialism is the foundation of the USA; rather the Constitution protects rights essential to allowing people to live free from tyranny and to pursue whatever path they choose.

Re:Hi got to be kidding

[MKalus]

Repealing the anti-murder laws is arguably a violation of the First Amendment. Neither extreme of anarchy or socialism is the foundation of the USA; rather the Constitution protects rights essential to allowing people to live free from tyranny and to pursue whatever path they choose.

And I guess giving everybody healthcare runs contrary to those goals?

Re:Hi got to be kidding

[pmz]

And I guess giving everybody healthcare runs contrary to those goals?

Yes, because national healthcare is fundamentally tyrannical in nature (excessive taxation, government-selected choices, inequality of application based on politically motivated criteria, likely free ride for politicians, etc.). National health care violates the Fourth Amendment, also, at a minimum (the government already has your health history...no warrant needed).

National healthcare creates an imbalance of power between the people and their government that is fundamentally incompatible with the intents of the founders of this country.

Re:Hi got to be kidding

[MKalus]

National healthcare creates an imbalance of power between the people and their government that is fundamentally incompatible with the intents of the founders of this country.

I can just see how it causes imbalance by keeping all those poor sobs at the bottom healthy, heck if they can't afford good medical care then why should I pay for it.

You are aware that Ideology relates to an ideal world, not to the real world?

The US isn't democratic, it isn't capitalistic either, neither was the soviet union ever a communist country.

Re:Hi got to be kidding

[MKalus]

And just in time I found this on the newswire.

The Software Value Proposition

[ozzee]

The future of the software market is clearly different to the market that existed before GNU&Co.

Customers make an investment in deploying software which ranges from equipment to training and maintenance infrastructure. These are the significant and notably "long term" investments.

This would imply that the wise customer would look not at the sticker price but also at the cost of the future investment when making a software purchase.

Unfortunately, given the behaviour of tech corporations over the last 20 years, this is the antithesis. Product development decisions are made based on profit margins and very seldom on helping the customer maintain their long term investments. Open source software however helps the customer maintain their investment by allowing them to take control of their future.

This is why Solaris will not succeed and has not succeeded in developing market share in the x86 market. Already is is obvious that Linux is a better product by virtually any meaningful measure.

Re:Solaris: Time machine to the 1980s

[bombadillo]

Most BSD's distro's are striped down in that fashion. It's just that linux includes every thing and the kitchen think. By the way, If you do install those GNU utilities make sure you put the GNU utilities behind the solaris utilities in the PATH. This is due to the fact that other solaris programs/scripts are sometimes dependent on the way the solaris utilities work.

My Momma?!?

[tizen]

I'm sure despite this little interlude, I'll still receive those flame-trolling comments. To that I say, If you have a problem with my choice in distribution, then feel free to run your own evaluation. Also, your momma is ugly. Seriously. UG-LY.

I feel like flaming him about his choice of distro, and also include a rebutal about his momma. He'll pay for his momma troll.

-tiz

Huh?

[Grendel Drago]

What are you talking about? Both benchmarks used gcc 3.3.1. Notice where it says "Compilation Time (GCC 3.3.1)" above the compile-time graph? Yes, he had to compile the same version of gcc on both machines, but he specifically doesn't include that task in the benchmarks.

--grendel drago

Re:Huh?

[Master Bait]

The benchmark was How Long Does it Take to Compile GCC 3.3.1.

Re:Huh?

[Ed Avis]

Er, from the article:

My first step was to use a common compiler, so I chose GNU GCC's latest, 3.3.1. For Solaris, I used the 2.95 GCC build I obtained from Sun's own freeware site (http://wwws.sun.com/software/solaris/freeware/). It compiled without any errors, and took 22 minutes and 26 seconds to compile.

Re:From the article....

[Anonymous Coward]

Hey dumbass mods, the parent is actually not offtopic .. that quote is actually in the article itself ahhaha..

Ehhhh ... wasn't that impressed with the review

[netglen]

Stepping outside of the Solaris --vs-- Linux article, I just didn't come away very satisified. It looks like the author only performed a few shoestring tests while I was expecting an exhausting barrage of tests. IMHO I wouldn't take anything to heart from reading this article. Maybe if the author goes back and expands the number of tests it would be an interestig article. Just look at all the comments from our fellow posters. There are so many people pointing out various issues that the tester neglected. I think the author should take all the feedback and perform a new test with an imporved set of criteria and hardware platforms.

3.3.1 vs. 2.9.5

[Anonymous Coward]

Compiling MySQL and the like with Solaris/2.9.5 and Linux/3.3.1 and eventually finding out that 2.9 ... err ... Solaris is faster.

B-r-a-v-o!

Taking into consideration how much more optimization - work gcc 3.3.1 does this is not _that_ surprising.

Re:SIR HAXALOT IS A KNOWN SPAMMER

[lanswitch]

Can you show us facts? They are more interesting than rants, thank you.

Re:Solaris: Time machine to the 1980s

[anomalous cohort]

It seems to me that most of GNU's file based utilities have some kind of -r recursive option. I could find no such option on Solaris. I realize that there is always a way around this with some combination of find, awk, and sh but the -r is really convenient for me.

Read again.

[Grendel Drago]

Did you notice the part where he says, "My first step was to use a common compiler"? Looks to me like he used the 2.95 to compile the 3.3.1 so that the same compiler would be running on both operating systems.

--grendel drago

Re:Read again.

[Anonymous Coward]

True, but the version of GCC that GCC is compiled in will have a noticeable effect on compile time (although not on anything else of note).

GCC 3.3.1 compiled in GCC 3.3.1 for both systems would've been fairer.

Re:Read again.

[Mr Z]

The final stage of bootstrapping GCC is to have GCC build itself. So, if you're running a fully installed GCC 3.3.1, it's actually a GCC built with GCC 3.3.1.

--Joe

1 vs. 2 CPU test?

[missing000]

It looks to me that he disabled the second CPU for the Solaris test, while he loaded an SMP kernel for the linux example.

I can't understand why you would expect a test like that not to favor the linux configuration. Given his results, I would have to say the Solaris machine was awesome.

Re:1 vs. 2 CPU test?

[tomhudson]

No, the article stated that since he was running dual cpus, he was probably in violation of Sun's license which allows for free non-commercial use on single-spu machines only. He then went on to state that, even if he WERE to disable the second cpu, that would be irrelevant to Sun's licensing, since the box would still be dual-cpu-capable (which is what Sun looks at - sort of like the RIAA's cd-burner math).

drivers

[AchmedHabib]

at least he was lucky there where drivers. Try installing it on real server hardware, then even FreeBSD has more drivers for stuff like RAID controllers etc. Look at the Hardware Compatibility List.
If you are planning on installing Solaris in your enterprise enviroment you have to buy hardware that will work with Solaris x86 rather than the hardware you normally use. But then again, in a enterprise enviroment, I guess one would choose the Sparc platform or buy the Sun Intel hardware.

Re:drivers

[rasjani]

Some shitface managed to sell a few intel (real intel, not just processor) rack servers to company where i work for and the machine came with scsi raid that was based on ami megatrends chipset. Ok, nice machines indeed at that time but the technical staff (that i wasnt part of at that time) insisted that we *must* use Solaris/x86 on the machines.

Well, all was good until the boxes where taken to *heavy* production: Raid drivers caused fs corruption and total deadlocks every 3-4 days. Feeling very superior our "solaris guys" insisted that we keep these things running, i guess it was about 6 months after initial move to production that they started to doupt the superioirity of the combination and about 50 on-site reboots..

Ofcourse, i understand that the problem was in the *hardware/driver* combination.. But the best part: One day before a critical product lunch, one of these machines was down again. I headed up to server room, took the beast offline, installed linux (some redhat, 7.3 i think) and installed the required java apps on the linux box. Needless to say, this box is running and it would have uptime over 400 days now (but it dont because we have had 2 server room relocations)

Ok, I do understand that the problem initially wasnt in actual operation system but the drivers and i cant put the blame on Solaris/x86 ... But i know this, since that move, we have replaced *loads* of Netra T1's with x86 servers and linux have had only minimal troubles (and only because of the laziness of techstaff. I think i should tattoo "Upgrades!" to my forehead)

Re:Solaris: Time machine to the 1980s

[Anonymous Coward]

Have you ever heard a sysadmin say "This box doesn't have enough memory for vim." or "vim is too confusing. Give me regular vi."?

Bullshit. Solaris' user environment is just primitive, crude, and OLD.

vim doesn't sell E10K servers, so Sun did'nt give a shit. They're going down the mainframe road -- the handful of actual system users have to bang 20 year old rocks together because what the Admin thinks of the system is immaterial.

Now that Sun is supposedly getting back into the workstation/desktop/small server market, they're going to have to upgrade their Unix utilities. It's a real world problem for them, because Linux is running right next door with modern utilities.

No surprise since you didn't RTFA!

[Anonymous Coward]

Solaris's strength lies in scalabilitly. I have a feeling Linux/SPARC will also beat Solaris/SPARC on a single CPU machine. But keep adding CPUs and watch Solaris scale (almost) linearly!!!

You didn't read the article did you?

The test box was an old VA Linux dual processor 600 MHz Pentium III machine that was bought used off of ebay. So that beats the "watch Solaris scale" theory.

Re:No surprise since you didn't RTFA!

[Anonymous Coward]

I'm sorry, but if you think that 2 processor performance shows evidence of scalability, then you've been badly advised. Or maybe you didn't RTFA?

"watch solaris scale" isn't a theory. It happens every day, running some of the biggest iron around.

Try it with 50 processors, or 106 processors in an E10K (on SPARC, I know, but this is what Solaris is designed for, and where it excels)...

This begs the question...

[Wakko Warner]

...who the hell still runs Solaris x86?

It's like Windows NT for Alpha... there's no point.

- A.P.

Re:Solaris: Time machine to the 1980s

[Neil Watson]

That pretty much sums up the world of commercial Unix. The commercial Unix companies had a lead of years (or even decades) on Windows and Linux. What did they do with that lead? Very little. Yes, they added hardware support and improved their kernels but, what about the utilities? What about the tools that Unix users need for everyday work?

Solaris V. Linux

[Greyfox]

After using apt, pkgadd will make you want to gnaw your arms off, even though it does have back-out capabilities. Sun should take that page from Debian and port apt over and start using it. They could add the missing capabilities from pkgadd easily enough.

Re:Solaris V. Linux

[bolthole]

already there, as far as getting free software goes.

http://www.blastwave.org/pkg-get

(think "apt-get")

Why Red Hat ? -- Bad info makes a bad article

[Anonymous Coward]

(too lazy to look up my account or make a new one, this is acdimalev, or tyln, for all of those who may have somehow seen my nick before)

Herein lies the problem, the assumption that all linux distributions are built equal is just plain wrong. To assume that a well tweaked Linux distribution runs no better than any other is one of the many faults of those new to Linux. While I've had many reports that 9 is a rather good release of RedHat I'd like to point out that there has been a long history of mismanaged installations, an excessive amount of resources wasted in error logs which literally fill pages with single events, and poorly optimized binaries.

I would have much rather seen SuSE used as it is currently the leader in simple, quick installation and management while maintaining several of the highest-running standards on the market.

In any case, I'd like to completely discredit the author in proposing that Mandrake, such the truly lacking and terribly broken dist (worse than RedHat) as it is, be a reasonable propal of alternate distribution for testing. Placing it on par with SuSE and Slackware was a mistake I hope this author never makes again.

And for those who still don't follow in my views, perhaps bringing to point that every test the author performs is based directly on system configuration. All tests are performed with ext3 in contrast to the much higher-performing reiser filesystem available to all Linux users of any decent Linux distribution, and these tests are filesystem access intensive. Also to note, the first half of comparisons is entirely distribution specific. Every distribution has it's own installation system, varying greatly in complexity and misfunction. Every distribution has it's own package management system, RedHat known for having one of the worst managed. And to round off, RedHat definitely has a record of the worst security management (across bigger distributions).

Please don't take this article seriously. Please don't promote articles like this. Misinformation is the greatest threat to all of Open Source. A person reading this article would hardly obtain any sort of scope of what is out there and would be tempted further to follow in the author's footsteps.

A public closed to being informed is a public closed to Open Source -- share the knowledge, share the freedom, promote the power of progress.

P.S. -- tar files are 'compile from source'

-- acdimalev

Re:Why Red Hat ? -- Bad info makes a bad article

[__aafutm5472]

I think the author went with Red Hat because it's one of the most widely used in business, which is also typically where Solaris is deployed. He didn't tweak either system at all, with the exception of installing the latest updates, because that's what your typical, non-Linux junkie sysadmin is going to do.

Having said that, I agree that not all distros are created equal, and that the results could have been far different (in either direction, mind -- he didn't tweak Solaris at all, either) had the author chosen a different distro.

But the author wasn't going for a "which seriously tuned system will win" type of comparison, just a "which system out of the box will win" type of comparison.

Generally accords with my experience

[BookRead]

I'd have to say that the article is in accord with my experience. I was all excited about SSH being in Solaris 9 but found the default install just too tedious to clean up from a security standpoint. I never really liked CDE, either. Ugly and insecure. In my last job, RedHat went on my 500mHz IBM desktop and the Ultra 5 gathered dust in the corner.

On the other hand, Solaris is really meant for large machines and the Sparc architecture. It works good there and still has the advantage in the mainframe-ish category. The gap is closing fast, though. Sun can continue the argue the "value" proposition for Solaris/Sparc but that's making sense in a smaller and smaller section of the market.

Once the 2.6 series gets going Linux is poised to push the proprietary Unices into extremely niche oriented markets. The resources that Open Source bring to bear on the software development problem is simply going to overwhelm any corporate attempt to keep up. The corporate choice is going to be lead or get out of way.

Re:The Linux cover

[psgalbraith]

The vast majority of your points are not worth a rebuttal. But here's a sample anyway:

Linux is an operating system. Like Windows, MS-DOS, OS/2, etc. There is no difference, in this sense, between Linux and other operating systems.

So WinXP is the same as DOS? Both can do the same thing and are as easy to use?

Linux is freely distributable, not free of charge.

Linux is not free, it's $4.99 at cheapbytes.com.

Because what one does verify, is that Linux is a hard-to-use operating system, at least in the install phase.

Step 1 - Buy Knoppix CD.

Step 2 - Insert on computer CDROM bay.

Step 3 - Turn on computer.

Step 4 - There is no step 4.

Security

[allenw]

I guess he doesn't know that Sun generally releases a T-Patch relatively quickly so that admins can get immediately relief while testing out the real patch.

Re:Security

[Anonymous Coward]

If by relatively quickly you mean 2 weeks... then you're right !

2 weeks is nothing when there are 2 big holes in your servers just sitting there....

Re:SPARC's Solaris?

[peachboy]

True, the hardware is called SPARC and the company is called Sun. However, the reason the submitter wrote "SPARC's Solaris" is because he/she was making a distiction between the SPARC and x86 versions, so they were correct.

A big boy

[Anonymous Coward]

So... Linux is a big boy now eh? Yesterday it was the BSDs now it is Solaris. With XP and AIX out of the picture months ago, what is next? .. ... ....
The world!

GOOD ONE.

[Anonymous Coward]

hey fuck-tits,

i'm holding a copy of x86 Solaris 9 in my hands right now.

since my hands are busy, why don't get your on my dick and hold that puppy while you suck it?

Article hit nail right on the head

[Anonymous Coward]

I use Solaris x86 and Linux at my job on a day to day basis for (oracle,informix,stronghold)and this guy has got it right. We only have 4way servers so we can't see the benefit of Suns OS scalability but in general the Linux boxes outperform the Solaris ones slightly in general interactive performance and alot in filesystem performance. This doesnt mean Solaris sucks, it has, what I think is good support, good drivers, good package system, good patch system and good reliability provide you have hardware on the sun compatibility list. And it's great to have Sun support behind you on your $30,000 servers to make the managers happy.

Haiku

[Anonymous Coward]

Solaris is great!
Or so the Slashdot troll said,
Linux like stale bread!

Re:Solaris: Time machine to the 1980s

[grigori]

Hmmm. On my Solaris box: bash-2.05 $ perl -v This is perl, v5.8.0 built for sun4-solaris. The software's there on the CDs. Just do it.

I did bother flaming a bit

[Anonymous Coward]

read up to the replies of "Why Red Hat ?"

PHP on Solaris isn't hard at all

[green pizza]

Ugh, why does everyone assume installing Apache, PHP, and MySQL is so hard on Solaris (or IRIX or AIX for that matter)?

Once you have your GNU environment configured, it's a simple matter of compiling. I haven't run into a snag doing this in over 3 years on three different commerical unices.

Here's a good link for the total newbie:
http://ampubsvc.com/~meljr/AMPS.html

I suppose you could also go to sunfreeware.com (or for IRIX, freeware.sgi.com), but learn to build the stuff yourself and you'll know what's going on, have the latest versions, and have way more flexibility. Isn't this why you're using u*nix anyway? For the flexibility? Don't let the lack of a precompiled ready-to-install package get in your way, you're not stuck in the Windows world anymore.
(end rant)

Re:PHP on Solaris isn't hard at all

[Anonymous Coward]

Ugh, why does everyone assume installing Apache, PHP, and MySQL is so hard on Solaris (or IRIX or AIX for that matter)?

Once you have your GNU environment configured, it's a simple matter of compiling.

So, you saying, after you get rid of of Sun's crappy tools and replace them with GNU's working tools, then it's easy. Yea, that's a good reason to use Solaris.

Solaris advantages.

[miguel]

Solaris does have a few areas where they have done a fantastic job.

For example, when it comes to debugging threaded applications, and having a reliable debugger, they beat us every single time. This is a mix of debugger support, kernel support, libraries support and god knows what else.

Their thread implementation is also very robust. I have no clue about their performance, but I know that you can depend on their implementations being robust. On Linux plenty of thread-related issues are still flaky (big progress being made there), but today, I really wish I had Solaris to debug a few problems.

And there are tons of other little things they get right. My suggestion is that we should focus on what is wrong in our platform, and focus on what is good in their platform, to find out what needs to be solved.

Miguel.

Re:Solaris advantages.

[LarryRiedel]

As a developer I find the Solaris documentation at docs.sun.com is usually significantly better than the stuff that comes with Linux, and the manual pages are vastly better, including information about when and why you would use a function, alternatives, examples, and pointers to other documents.

To me documentation is a big part of any platform. I think that is one thing .NET has going for it too.

Larry

Re:Solaris advantages.

[Anonymous Coward]

> For example, when it comes to debugging threaded applications, and having a reliable debugger, they beat us every single time.

I agree. Also it's great that they ship with kernel debugging stuff (adb -k, kadb, crash, ...). Most of the same functionality is available for linux but not in the mainline kernel (for really irritating reasons) so it's never there when you need it :-(

For userland stuff I usually use gdb even on Solaris, but often times I'm forced to go back to using adb when dealing with threaded apps.

> On Linux plenty of thread-related issues are still flaky (big progress being made there)

Yes, I believe that the 2.6 kernel and the latest libc work effectively fixes this.

The other thing I like about working with Solaris is the nice set of performance monitoring tools (iostat, cputrack, cpustat, mpstat, prstat) available in the stock distribution. In my work I use these almost daily. I really miss having that level of info at my fingertips when I'm working on the linux machines.

But yeah, for most purposes I like developing on and deplying linux far more. All in all I think RH9 is a better OS than Sol9. (And I'm *not* some "linux wanker" - I've been working with Solaris since 2.1 and even SunOS 4 before that)

Re:Solaris advantages.

[Brendan Byrd]

Given that most of the commands in Solaris lack the functionality of most of the commands in Linux, it would not surprise me if Solaris was able to document all TEN of their command switches for "ls". Linux is a little bit more complex, thus documentation is a bigger deal. However, given that it works on more hardware and has more software/commands for it, the documentation for Linux isn't all that bad. "man [command]" helps me for most of my needs, and "apropos [subject]" helps for the rest.

Re:Solaris advantages.

[thanasakis]

Solaris does have a few areas where they have done a fantastic job.
Sun also has the edge regarding documentation, IMHO. Besides docs.sun.com (which is an excellent piece of information) the man pages seem to be more complete in certain areas. I wouldn't go as far as to get bitter about many great gnu/linux apps which totaly miss even the modest form of documentation, but there is a distinct difference.

we should focus on what is wrong in our platform, and focus on what is good in their platform
Funny. Would you please care to define "we" in contrast to "their". Or do you mean "we" as in gnu/linux, in contrast to "their" as in solaris. Some people, including me, believe that there are more common points between the gnu/linux movement and solaris in general. Namely:
1)Sun has donated the NFS. Everybody including gnu/linux uses it nowadays
2)Everybody uses mozilla right? Sun has donated heavily on mozilla. How about that.
3)Openoffice. Largely thanks to Sun, we are able to write documents,presentations etc in a product comparable to MSOffice that can run on GNU/linux.
4)GNOME. I bet a lot of gnu/linux users use gnome. Guess what! Sun has donated large sums to gnome too! If fact they are phasing out CDE in favor of gnome.
5) Sun is probably one of the few companies that does not lock you in her hardware/software/OS. They
remain largely a traditional unix OS and that's why many many people are able jump from one to the other instantly.

I could go on but you get the idea. If some of the /. crown regard Sun as the enemy, fine. But I do not think that Sun regards GNU/linux as its enemy. Or at least even if they do regard it as enemy, they haven't done anything serious yet. Compared to other firms (you know who) they have practicaly endorsed it.

Solaris/Inetl remains still a viable solution for many tasks IMHO. The fact alone that almost 95% of open sourced applications can be compiled and used under it makes it a vary viable choice. A normal GNU environment including gcc,gdb and other tools is available for solaris intel. So why would the gnu/linux community regard solaris and sun as its enemy? I for one, use solaris at work and gnu/linux at home and you know what? I am happy that they are more like than unlike so that I have choice to use whichever I need to. Plus, when multi-core CPU's hit the market, solaris might have an edge depending on Sun's expertice in the area of multiprocessor systems.

Now if Sun could release the source of the solaris kernel, a new GNU OS could be build, something like GNU/solaris. Oh well, I can go back to sleep now.

Re:Solaris advantages.

[a1291762]

My suggestion is that we should focus on what is wrong in our platform, and focus on what is good in their platform, to find out what needs to be solved.

Well, for a start, how about not changing major interfaces (kernel and userland) every release? Solaris apps from long ago continue to work with little or no modifications on the current hardware/OS. Linux can't claim that. It's the distros that are at fault with userland of course, but there's no reason why version x+1 of a program/library/whatever needs to be incompatible with version x.

It's all about your focus. Sun is like IBM in that regard. They want customers to migrate their stuff from older systems to newer ones. You shouldn't need to rewrite everything to do that. Unforunately, the Linux world seems to have copied Microsoft's philosophy of "change the interface every release so we're a moving target". Serious companies don't bother with that.

Link

Re:Solaris: Time machine to the 1980s

[pmz]

it doesn't ship with perl (last time I checked)

When was this, 1993?!? Get your facts straight before posting (like that stops anyone on Slashdot, anyway).

Re:Solaris: Time machine to the 1980s

[pmz]

It seems to me that most of GNU's file based utilities have some kind of -r recursive option.

GNU's not UNIX. The proper way to do recursive operations in UNIX is to either embed it into a find command line or to do a for-loop in the Bourne or C shell. Coding a -r operation into each individual command is a good example of how GNU went awry.

Re:Solaris: Time machine to the 1980s

[N1KO]

These are the vi packages on available on my system:

vi (3.7-r4): The original VI package
bvi (1.3.1): display-oriented editor for binary files, based on the vi texteditor
nvi (1.81.5-r1): Vi clone
vim (6.2-r3): Vi IMproved!
gvim (6.2-r3): Graphical Vim
kvim (6.2.14): KDE editor based on vim
vile (9.3h): VI Like Emacs -- yet another full-featured vi clone
elvis (2.1.4-r1): A vi/ex clone
xvile (9.3h): VI Like Emacs -- yet another full-featured vi clone
viper (1.35): VI emulation support for Emacs.

Vim wasn't installed by default, I could have installed any of the other packages if I wanted.

Re:Solaris: Time machine to the 1980s

[Xrc65kl]

Have you ever checked out the /opt/sfw directory on you sun machines?
A solaris installation from the normal Sun disks puts all sorts of gnu utilities there. Vim too.

Re:Solaris: Time machine to the 1980s

[shaitand]

yes it's true, how horrid to not require me to script to perform an operation recursively and making it a simple switch instead (while not removing the capability to use the script). yeah your right, they really fucked up big time with that. And making these additional features open and free, instead of closed, prorietary and expensive. How dare they?

-1 Troll

[Anonymous Coward]

This is "insightful" because it is currently in vogue to bash Sun on slashdot.

It must be a long time since you looked as Solaris.

Solaris ships with PERL as a supported part of the OS bundle, and also GNOME and bash.

In the Solaris 9 (and 8) media kit is the Solaris Freeware (sic) Companion CD which contains such things as Vim, Python, gcc (2.95.3 and 3.x), KDE, XFce, ethereal, PHP, GIMP, ImageMagick, CVS, plus loads of other useful stff.

Re:Solaris: Time machine to the 1980s

[kindbud]

... and it doesn't ship with perl (last time I checked).

Then you must have last checked around 1999 or so. Perl has been standard since Solaris 8. Several new system utilities require it.

someone please

[Anonymous Coward]

mod parent 'troll'

Noone supports Red Hat 9

[christophersaul]

It would be interesting to see a comparison with the paid for versions of Red Hat, as opposed to the 'support yourself' versions - the supportability of Solaris is important for many customers. That doesn't necessarily just mean bug fixes - having someone to call is important, particularly where it's hard to get decent sysadmins. Red Hat 9 is great, but Solaris offers predictability, support and so on and so forth.

Solaris x86 is the best fit for several problems..

[emil]

Why would you pick it?

• Threads. Solaris has the best support for threads on x86.
• UNIX98. If you need an OS with this type of certification on x86, Solaris is the only game in town.
• NFS. While I have no quantitative data, I would bet you $100 that Solaris x86 NFS stomps Linux.

There are probably lots of other reasons. Solaris is by no means perfect, but it does have its strengths.

Re:Solaris x86 is the best fit for several problem

[GoNINzo]

Sun's NFS is only faster if you're dealing entirely with other Sun boxes, but I've only tested it on Sparc. I would actually put money on Linux for that. Or, you could look at NetAPP if you want real speed.

You're right about UNIX98 though. heh

Re:Solaris x86 is the best fit for several problem

[Anonymous Coward]

Linux NFS is only useful if you're deailing with other Linux boxes. Otherwise, it's got quite a lot wrong with it.

pure rubbish

[jeramybsmith]

The solaris 9 installer stinks,
the patch system is horrible,
the package system is far behind dpkg or rpm

Seriously, this author has a serious hard-on for solaris and it shows. Solaris is stuck in the year 1997.

This isn't a troll, this is just the way it is.

Re:pure rubbish

[swordgeek]

THAT's a hard-on for Solaris? Geez, I'd hate to see if he wasn't so 'biased.'

I don't know when the last time you installed or worked with Solaris was, but:

The installer is straightforward and works really well.

The patch system is ahead of anything on Linux.

The packaging is a small step behind dpkg, but is also much easier to use. Both of them are at least three orders of magnitude better than RPM.

Solaris is also likely slower overall than Linux on x86, more stable, less hardware-universal, and able to run fewer apps. Some good, some bad, some different. If you can't see that, then YOU are apparently stuck in 1997.

Re:pure rubbish

[chegosaurus]

The Solaris installer is ridiculously simple, and it has Jumpstart, which is the best installation system going. (The Webstart installer stinks though - no one could argue with that.)

The patch system is 100% fine, especailly if you're using PatchPro. Patches could install quicker, but it's not a problem really.

The package system isn't great, but it does the job most people require of it. Solaris is a server OS - you don't generally add much packaged software, and there often aren't many dependencies to fill. It's also an OS often maintained by the kind of admins with the smarts to build their own binaries targeted for their own systems.

Hardcore backwards compatability

[devphil]

Most slashdotters won't understand or agree with this, but the large bulk of Sun's customers appreciate the fact that command-line options do not mutate over time, that the default behavior of the -foo switch is now reversed, etc, etc. The GNU coreutils maintainer has been busily ripping out all kinds of traditional functionality in the name of POSIX standardization, which would normally be a good thing if he hadn't gone way too far. (I don't give a fuck if "uniq" and "head -1" aren't full POSIX, they're in my scripts, they're in my head, and they're staying there.)

If Sun tried to make as many incompatible changes to their core utilities as the GNU utils does, somewhere upwards of 80% of the customers would just walk away.

Yes, I install GNU coreutils and all kinds of happy stuff (like a decent shell) as soon as I open up a Sun box. But I leave their versions in place so that old PATHs still get the behavior they expect to find. Everyone here loves the cutting edge, and loves to cut down anyone using version ($latest-1). Sun's primary customers aren't like that. They want stability in the core utilities across years, not new features every few weeks, or even months.

(Yes, the last 3 or 4 versions of Solaris have all shipped with Perl. It's a slightly older, stable version of Perl. There's a bunch of stuff on the freeware companion CD too, as well as sunfreeware.com. Those who want a stable Solaris get it by default. Those who want bleeding-edge tools can easily download the packages.)

Re:Hardcore backwards compatability

[haggar]

like a decent shell

I really don't understand what you or anyone else criticizing Solaris in this way, is missing when it comes to shells.

Solaris comes installed with bash, ksh and a statically linked bourne shell sh, useful in single-user mode when the root filesys is all you have. These are all decent shells, each has it's strenghts and weaknesses. Want POSIX, loads of builtins, powerful mathematical and string processing and displaying options for your scripts, chose ksh. Want a very friendly shell for your dayly admining and navigation in your Solaris servers, chose bash. And when the shit hits the fan, then you have sh - much better than nothing, you'll have to reckon.

Re:Hardcore backwards compatability

[devphil]

That's not what I'm criticizing. In fact, I strongly agree with your right-shell-for-the-right-job attitude. (Sun could ship a newer bash, but enh, whatever.)

The criticism is that /bin/sh is a flaming piece of shit. Not "traditional Bourne shell" in general, but specifically Solaris /bin/sh. It is full of quirks, bugs, and incompatabilities. And they will not change it, for the same hardcore backwards compat reasons I mentioned before. Somewhere there may be a customer whose ancient entangled shell scripts unwittingly depend on funky Sun behavior, and Sun will not break it for them.

The result is that if your correctly-written /bin/sh script works everywhere else, it may not work on Solaris, or it may only work some if the time. The GCC project found this out the hard way; occasionally Solaris /bin/sh will just fail at random times, on identical scripts. (See their FAQ.) If you have /usr/xpg4/bin before /usr/bin in your PATH, then "sh" actually gets you ksh (well, ksh88), but who starts their Bourne shell scripts with "#!/usr/bin/env sh" ?

Re:Hardcore backwards compatability

[haggar]

First of all, /bin/sh is a statically linked minimalistic shell, used in emergencies (single user mode, low memory etc.) and it's conveniently located in the root fs. (nowadays the defauls Solaris install places /usr in the root fs by default, too (which I consider a sane decision) but that was not the case previously in history). That shell is not supposed to be used for complex scripts, or scripting at all. It's a less-than-100k binary that does the job in such emergency situations.

Second, if you want your scripts to be portable, you don't write for Bourne (or even bash) shells, but for a POSIX compliant shell, such as ksh. That's why all our scripts have a header that finds the system on which it's run, and then relaunches the script with the POSIX shell for that system. Obviously this is not the way you do it, so if you are curious about that header, I can post it for you.

Re:Hardcore backwards compatability

[devphil]

First of all, /bin/sh is a statically linked minimalistic shell,

No it isn't. You're thinking of /sbin/sh, which is what root's shell is set to.

And it's not exactly minimalist, either. /bin/sh has nearly all the POSIX features, they're just buggy as hell.

As far as scripting goes, every Bourne script I've seen starts off "#!/bin/sh", under the (reasonable) assumption that the shell there does not, in fact, suck ass. The C/POSIX interface bindings even specify that /bin/sh is the location of the POSIX shell; another place where Solaris falls down in the name of backwards compatability.

For re-execing shells, I typically use the header found in the Goat Book, or variants thereon. Usually I would do that in order to use ksh or bash features in a startup script in /etc/init.d, which Solaris unconditionally runs using /sbin/sh.

Re:Hardcore backwards compatability

[haggar]

You are right about /sbin/sh, and I am wrong. I still feel I am on the right side, some way or another, but right now I can't quite point it out why :o))))

Thanks for the link to the book. I found the shell re-executor. What we use is a bit more elegant and less obvious, but I am afraid I can't post it on slashdot as it's part of our products, and I work for a paranoid and huge company.

Re:Hardcore backwards compatability

[devphil]

Yes, I think we're in violent agreement. :-)

See Register article on new TCP/IP stack

[grigori]

Hey this just came out at http://www.theregister.co.uk/content/61/33440.html
"The new TCP/IP stack - code-named Fire Engine - has 10 gigabit and 100 gigabit Ethernet networks in mind." Available for testing download now

Sun is quoted:

"we did work to efficiently handle many NICs, so combinations of NICs and CPUs scale. The upshot of all this stuff is that customers will notice quite measurable differences in latency and bandwidth improvements in networking on existing machines. It shows up in internal benchmarks, and it shows up in real-world workloads from the smallest uniprocessor on up. Also we focused on CPU utilization. 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."

4 processors max for x86 Solaris

[sacrilicious]

Who the heck runs Solaris on a crappy dual-processor machine? Solaris doesn't really even begin to show benefits until it gets at least 8 processors. It just keeps going up, and up, and up.

According to the article, Solaris for x86 runs on a maximum of 4 processors.

Re:4 processors max for x86 Solaris

[DNS-and-BIND]

Solaris/SPARC is the real version. Solaris/Intel is the toy version.

Re:4 processors max for x86 Solaris

[Anonymous Coward]

Solaris/anything is the dying version.

Re:Solaris: Time machine to the 1980s

[mihalis]

Running Solaris on X86 is like going back in time to the 1980s with all the ancient Unix utilities -- [SNIP] it doesn't ship with perl (last time I checked).

So, it's been a long time since you checked then, eh?

This depends greatly

[mindstrm]

on what you perceive as the role of the sysadmin.

In some areas, the developers know far more about the target systems than the sysadmins do... (say, compiler design, OS design, stuff like that)

In other areas, the sysadmin has a far better understanding of the hardware than the develoeprs (office applications, productivity software, most mass-market stuff). I can't count the number of times when I've asked some developers of software we bought how something interfaces with the hardware, what protocol it uses on the network, somethingl ike that, adn the answer from the head developer was "We don't know; we just used this library"

Be careful!

[pr0ntab]

Which is bizarre, because I've discovered things like the ypserver portion of NIS works much better if you use gmake to update instead of plain-jane make.

Fresh install, ypinit -m, then make, and boom! errors. No update pushed to slaves.

So you have to go in and futz with some targets with missing dependancies.

OTH, use gmake, and it ignores those, and prints a warning. Do the script maintainers have their paths wrong??? ;-)

Re:Solaris: Time machine to the 1980s

[Brendan Byrd]

I was able to replace most of the stuff, such as a "ls" command with color. However, certain utils, such as "ps", I couldn't find replacements for.

In any case, that one Sun box kicked the bucket, so I'm not to worried about it anymore, but I was busy trying to transform Solaris into GNU/Solaris.

Next time though..

[Mr2cents]

He should normalize the graphs. I mean, on one graph linux is in purple and solaris in blue, on the next it is reversed. This is very confusing and unnessecary.

Thank god.

[pr0ntab]

It's about time.

And I hope one of the tasks for 2.7 is an infrastructure change to handle drop-in TOE accelerators. Such a restructuring can probably expose those same CPU savings as a side effect of such posturing (and giving NIC cards more chances to offload work).

Re:Solaris: Time machine to the 1980s

[Brendan Byrd]

Oh, sorry. I didn't know that the purpose of GNU utilities were to make life HARDER. I guess I should not even bother with commands, and just type in the machine code by hand.

And I suppose that when I type "killall" without any parameters, it should kill every process on the machine, instead of giving me a help screen with the list of options. (Recounting an AIX experience...)

complain away.

[Anonymous Coward]

how can he do this?
this is nuts. he's an idiot.
he doesn't know anything about linux.
he doesn't know anything about solaris.
why didn't he optimize solaris better.
why didn't he optimize linux better.
that software is irrelevant.
this test means nothing.
he shouldn't have posted it at all.
there. do you feel better?

Re:Solaris: Time machine to the 1980s

[jedidiah]

Bill Joy doesn't even use vi anymore.

The rest of you should take that as a hint.

At least RMS eats his own dog food (emacs).

Where can I score a copy?

[Anonymous Coward]

I'm an O/S super freak. Anyone know where I can get a copy of Solaris x86 to try out? I just want to install it, play around till I'm bored with it (an hour or a month), then delete it. The usual.

Re:Solaris: Time machine to the 1980s

[pmz]

The point is one of simplicity and robustness. Writing single-purpose tools each with limited scope shrinks the debugging burden when problems occur. Using 'find' to handle the recursion is a no-brainer.

Pure UNIX is good software engineering. Feature bloat in the name of convenience is Microsoft engineering. You might as well use Windows.

Re:Solaris x86 is the best fit for several problem

[Brendan Byrd]

UNIX98? What's that? The competitor to Windows98?

Solaris on x86

[agentk]

Why the heck would you run Solaris on an x86 instead of Sparc? Linux was written for the x86. Solaris was written for the Sparc. Each uses different approaches depending on what the architecture offers or doesn't offer. So why bother with this kind of comparison?

Yeah, the great IBM.

[Anonymous Coward]

Who cares that they are one of the worst corporations still around? Having done everything from helping design and impliment systems to track and sort jews for the nazis to giving us microsoft. But hey, who cares about little things like that, they said "linux" for their own self-serving reasons, so they are the fucking greatest aren't they? If you really cared about linux you would support a smaller vendor that actually gives 2 shits about linux, instead of climbing inside IBM's ass hoping you can smell a little more like *their* shit.

Re:Solaris: Time machine to the 1980s

[Brendan Byrd]

I believe that a command has failed when you have to resort to commands such as xargs. Any sort of command that has a filename as an input should also have a recursive option. It's not bloat; it's used all of the time. I use "grep -R" all of the time (especially in the /etc directory). The fact that I could use "grep `find .`" is another way of doing things, which is okay. There's nothing wrong with having a couple of different way of doing things.

If that sort of functionally was implemented into one library, it wouldn't take up all that much space, anyway. I use Linux/*NIX because it's designed to be a usable console, versus Windows, which is a GUI, and DOS, which is a stripped-down and unusable console. Bloat doesn't become an issue unless your minimum requirements continue to get raised, but I can still install Linux on a 386 if I need to. I can't say the same thing about Windows XP.

My reason for using solaris x86

[Phibz]

Although i am a Solaris admin professionally and am a bit biased, my main reason for using Solaris x86 on one of my machine's is for its nfs preformance. Its preformed considerably better than linux at the task. Even under very heavy load it hasn't broken. I was surprised when i loaded 9 and found it even supported my 64bit pci adaptec ultra-160 scsi card.

phibz

Re:Solaris: Time machine to the 1980s

[Anonymous Coward]

You are an idiot. Solaris x86 has been distributed with perl since 5.8 and all the other software you mentioned is distributed in a very easy to use package format on http://www.sunfreeware.com for free! So shut up!

Good article, slightly anti Solaris

[mnmn]

The conclusion was a bit opinionated and not as analytical as I'd like, but its true, Linux does trample all over Solaris both on x86, and with Solaris running on a comparable sparc platform.

Now the whole focus of Solaris has been in a different direction. Being UNIX, Solaris is a very standard market OS, it is THE standard and most commonly used UNIX out there. No other UNIX has such a huge codebase. You download oracle or websphere, and it installs without any package-type or library problems. Linux comes in too diverse a flavor to allow that.

Solaris is also poised to be more stable and robust than fast. Linux 2.6.0-test8 with a preemptive kernel and XFS filesystem runs nicely. But Solaris runs robustly. They're both miles away from Microsoft in stability, but you get issues with the thread libraries in Linux more often than in Solaris. In Linux, theres a large set of drivers that are EXPERIMENTAL, and for many things, you have to get the packages from various places (think ATM networking or ipsec VPN) while theyre already sitting there in Solaris.

In that sense, Solaris is more like FreeBSD, and comparisons with FreeBSD would be more interesting. What I'd like to see however, is a large shootout between Linux 2.4, 2.6, Free/Net/OpenBSD, AIX4.3/5.1, Solaris9 (sparc64), IRIX, BeOS comparing filesystem performance, threads, IO, network throughput, number of packages available and the likes.

gcc 2.95 on sun vs. 3.31 on Linux

[Anonymous Coward]

This guy doesn't seem to be very precise in his comparison. To compare compile times between different platforms with different compilers is not very expressive. I think gcc 3.3 takes at least 10% more compile time than gcc 2.9

Re:Solaris: Time machine to the 1980s

[Anonymous Coward]

Pipes? Hello?

Two crucial issues left undiscussed

[swordgeek]

The article missed out on the two absolutely essential items which are probably Sun's strongest points, and Linux's weakest.

1) Documentation.
2) Versionitis.

Sun's documentation is wonderful. In addition to man pages which are remarkably up to date (a far cry from the bad old days), there's docs.sun.com which has all of the man pages PLUS professionally written and reviewed manuals. Linux has a mishmash of docs which are often out-of-date, frequently poorly written, incomplete, and sometimes just plain wrong. (Exceptions exist of course--iptables is wonderful.)

Almost the only time specific versions of packages are required on Solaris is when you're doing something with third-party software. The Sun packages (and nearly all stuff from sunfreeware, for that matter) go in and Just Work. Every time I get a package for Linux, it seems like I need to update the version of something else.

Performance vs. stability? That's a minor issue compared to a lack of formal and complete documentation.

Re:Two crucial issues left undiscussed

[Anonymous Coward]

Linux has a mishmash of docs which are often out-of-date, frequently poorly written, incomplete, and sometimes just plain wrong.

Thats not necessary true. During a usenet NT/Linux flamewar I heard a rumor of an announcement on an IRC channel of an incredible set of Linux documentation. To get the documentation you just have to connect to a gopher site with a copy of the announcement in sgml format. The announcement contains a link to the "el Dorado" Linux documentation collection. The el Dorado site has the html and pdf files from a high end technical publisher that went bankrupt while they were preparing a series of Linux e-books. The liquidators didn't think that there was any value in "free" software, so they released the books into the public domain! The books are awesome, but for some people they will have two drawbacks. First, some people might need to "patch" the books for their own use since they are written in literate, technical Urdu and are just ripe for translating. Second, the el Dorado web site can only be reached via a slip connection established by a cron job which runs on certain Julian calendar dates to avoid being slashdotted when they make announcements. I'm kind of looking forward to those documents getting wider circulation since they may be the only way to clear up some of the mysteries found in the GNU/Linux documentation. I'm not sure, but I think that more than one man page/info doc has Cthulu listed as its maintainer.

Any way, once documentation of this quality starts making the rounds, IBM, Sun and HP's days are numbered!

62532-02943

My guess would be

[hayden]

When Sun stops supporting SCO and trying to issue smackdown to linux at every oppertunity.

Re:Solaris: Time machine to the 1980s

[Anonymous Coward]

It's called "killall" for a reason foolio.

x86 Disconituined

[attobyte]

Wasn't Sun going to stop Solaris on the x86 with 9?

Re:x86 Disconituined

[chegosaurus]

No. They were going to stop it with 8. There was an early beta of 9, then it was pulled. It was probably the best part of a year before it got resurrected.

Re:Solaris: Time machine to the 1980s

[Anonymous Coward]

Allow me to introduce you to certified Unix(R) products. Since you probably aren't familiar with them, branded Unix(R) products meet a strict set of standards to ensure compatibility. Linux might some day qualify to be Unix(R), but not today. Until then it is just a "unix-like" operating system. If Linux ever becomes a certified Unix(R) operating system, it will look like the other certified Unix(R) operating systems. That means one of two things: either pack for bags for the "80s", as you refer to the standard, or the Unix(R) standard breaks with existing interfaces and functionality. Which do you think will happen? Personally I'm not betting on there being a whole gnu standard, nudge nudge.

Re:Solaris x86 is the best fit for several problem

[calidoscope]

* NFS. While I have no quantitative data, I would bet you $100 that Solaris x86 NFS stomps Linux.

Kind of funny that a reasonably well documented protocol has not been properly implemented on Linux - then again, the philosophy underlying NIS/NFS is somewhat foreign to Linux.

There are a few things in userland that work better in Solaris than Linux. Xsun has true PostScript code and will correctly render files that don't render on Linux. sdtimage has a much more intuitive interface for cropping an printing than any of the common Linux picture utils. My experience with dtterm has been happier than any of the terms on Linux.

I also have a bone to pick with the author of the review - /opt is a Good Thing. Putting all of the non-system applications in /usr/local is simply asking for an administrative nightmare. For people complaining about how dated CDE is, putting stuff in /usr/local is even worse.

Great things about Solaris

[chegosaurus]

Forte - most open source apps built with Forte on Solaris considerably outperform the same apps built with GCC. Sun's C compilers are the shit, and that's before you even get into debugging with them.

Docs - I get so frustrated with sucky docs on Linux. (Or, more particularly, no docs at all.) And don't give me that "you have the source, that's the best doc of all" crap.

Stability and predictability/ I *like* the fact that most of Solaris hasn't changed since 2.4. I know where I stand on any machine. I still use the old Sun sed/awk/ksh/etc etc because I know that whatever Solaris I have to work on, they're always there.

Okay, so the open source apps Sun bundle are always a few versions behind the current release. So what? I know they've been tested above and beyond what the authors tested, and if it's still a problem I can build the new version myself. With Forte.

Support - far and away the best I ever used. Partly a result, I guess, of having the same people make the hardware and the OS. I never understood how it's bad that MS make an OS for Intel h/w, but it's no problem to make Linux for AMDs.

Re:Solaris: Time machine to the 1980s

[salimma]

Bill Joy does not work for Sun anymore too. :)

Re:Solaris: Time machine to the 1980s

[Random832]

that's hardly the point... for those of you not keeping track, bill joy _wrote_ vi, originally.

Re:Solaris: Time machine to the 1980s

[salimma]

Ah! Now that makes more sense. Though of course he would now be using Vim :)