Solaris vs Linux Continues

raffe writes "Solaris Kernel Developer Eric Schrock is bloging more about the Solaris vs. Linux issue and linux kernel moneky Greg is answering on his blog. Eric's first part is is also still up and Greg's answer " Another reader also submitted reviews of the Linux desktop vs. Solaris 9. User reviews are welcome; please note that ITMJ is part of OSTG like Slashdot.

Re:Kernel Recompile

[StuartFreeman]

apt-get install kernel-image-X.X.XX

Re:Kernel Recompile

[Nos.]

I really don't think kernel recompiling is the biggest thing keeping Linux from dominating any market. Ease of use is a big thing. Another is simply the myth that OSS is unsupported and/or unreliable. You can point to a thousand studies showing Linux is as good as (or better) than alternatives, but that won't change some peoples minds.

Re:Kernel Recompile

[chez69]

i've used linux for exclusively for over 7 years at home and I've never recompiled my kernel.

properly packaged distros usually do not require a kernel compile.

Linux versus X

[Ruie]

Just to cover various "Linux versus X" topics, here are some links, obtained by Googling, without RTFA:

• Linux Versus NT [wired.com]
• Linux versus FreeBSD [prohosting.com]
• Linux versus TwinView Nvidia GForce4 MX 4000 [alufis35.uv.es] (ok, it's a bizarre one, but we are being thorough :)
• Linux versus MacOS X Server [prohosting.com]
• Linux Versus On Time RTOS-32 for Real-Time Embedded Systems [on-time.com]

And, most importantly: Linux Versus Linux [currents.net]. (No you can't actually read it..)

Ok, this was the first page.. I got bored copy'n'pasting afterward.

What is up with you armchair kernel hackers?

[808140]

I mean seriously. We have a debate about the relative merits of Solaris and Linux, and you come out and say, "LOL no context haX0rs@!!~ OMFG linux is so wei faster than Slowaris lol!"

I mean, did you even read his blog entry? I know, I know, this is Slashdot. But come on. He isn't comparing Linux and Solaris as gaming platforms. Yeah, your FPS for Doom 3 is probably faster on Linux (LOL d00d don't you know Doom 3 doeznt run on Slowaris haha you fail it!) but what he's talking about is no downtime, ever.

He's talking about kernel debug utilities. About hardware hotswapping. About being up 24x7x365 doing 1000s of database transactions per minute. We aren't talking about your mom's basement here, with your little network, or even the nice little RAID setup you have going at work that saved your employer a pretty penny. We're talking about big iron. Speed is not the issue here; reliability is. One of the reasons Solaris is slower than Linux is because it checks everything. It is one extremely anal system, and it never ever goes down.

Now, I'm a big Linux fan (typing this on my Debian box), but no one who has seriously admined Solaris boxes can say that the two are even remotely equal on big servers. No contest indeed; Solaris kicks the shit out of Linux.

I don't think this will be the case forever. Unlike the anal blogger referenced in the writeup, I think Linux is catching up faster than Solaris is improving. While he makes good points about Linux's lack of sysadmin accessible kernel debugging tools, traceability, etc, people attempting to sell Linux to big vendors will provide those tools.

But Linux isn't ready for the big iron machines Solaris dominates yet. Don't say IBM, please. IBM runs multitudes of instances of the Linux kernel in parallel on their machines, so that if one fails, it doesn't take the whole system down. Those big iron Sun machines run one kernel, baby. Just one.

I tell you, if they open source Solaris (yeah right) we're going to be looking at some pretty amazing code. Some of the best hackers ever have hacked that thing.

Text of Greg KH's post (in case of /.'ing)

[Anonymous Coward]

I was glad to see that Eric took the time to address my previous rebuttal to his previous comments. I welcome good technical discussions like this, in the open, without rude flames by anyone. It's fun, and lots of people get to understand things a bit better about the topic

That being said, I'd first like to address his closing comment, which was regarding my comment about Linux not going anywhere:

For some reason, all Linux advocates have an "us or them" philosophy. In the end, we have exactly what I said at the beginning of my first post. Solaris and Linux have different goals and different philosophies. Solaris is better at many things. Linux is better at many things.

I agree completely. I wasn't trying to put up any "us vs. them" type attitude, I was merely trying to explain in my message the reasons why the Linux kernel has or does not have those different features that Eric was discussing. My comment at the end was a bit glib, I agree, I was merely trying to state that Linux isn't going anywhere, and will welcome all Sun users and developers if they decide that Linux will work for them.

Ok, on to the technical stuff:

First off, thanks for giving specifics about your points of reliability, serviceability, observability, and resource management. Let's address these points.

• Reliability - Of course reliability is more than "better than Windows." Geesh, what a low bar to shoot for these days. Linux had better be able to handle hardware failures where ever possible, when ever it can be detected. Ah, that last part is the biggest issue. Linux most often runs on hardware where such errors can not be detected, as we run on a zillion different platforms (although not as many as NetBSD). For systems that we can detect these kinds of errors, we do (like PCI error reporting on the PPC64 platforms for example.) The hardware that Solaris usually runs on also has that kind of error reporting capabilities, and so the OS takes advantage of it. So Linux and Solaris are pretty equal here. As for the claims that the ZFS people are stating, I think that Linux filesystems like Lustre and SSD do pretty much the same thing (automatic error correction for large collection of disks all without the application needing to fix it up.)
• Serviceability - Sure, things go wrong all the time. That's why enterprise distros add the crash dump, kernel debuggers, and dprobes code to their kernels in order to be able to help service their customers. Nothing different from Solaris there (although you mentioning the ability to have a firmware dump of hardware errors is pretty cool, but again, that's a hardware feature, not an OS one.)
• Observability - DTrace does sound like the all-singing, all-dancing solution to everything that a kernel could possibly report to a user. And if so, I commend you all for creating such a wonderful tool. As for Linux, if you want much the same functionality, use the LTT code, or the dprobes code. Again, many enterprise related Linux distros ship their kernels with these features added to them, as their customers ask for it.
• Resource Management - That sounds pretty much exactly what the CKRM project does for the Linux kernel. Again, enterprise distros ship it, so their customers can have it. And this feature is getting fixed up to be acceptable for the mainline kernel, and will probably get merged into it within a year or so (but again, if you want that option, it's available to you.)

As for the comment about Solaris having these features "more polished" than Linux's, I will not disagree. But they are getting better over time, as companies realize they want these features in Linux, and address any shortcomings that these features may have.

Binary compatibility. You state:

We have customers paying tens of millions of dollars precisely because we claim backwards compatibility.

You have customers paying that much money f

Re:Question for anyone...

[mr_majestyk]

Can anyone cite a real life example where Solaris was used in place of linux on a new project for a valid reason?

Here's one [sartryck.idg.se].

The reasons? Linux couldn't handle emergencies, and wasn't always available.

I know.

[Ayanami Rei]

That still bothers the FUCK out of me.
I mean, it's easier to set the terminal speed of the real serial port in the firmware to a decent speed, and use that over a minicom session to a nearby linux box. Set your consoles to ttya, boys; never mind that extra $500 Radeon 7000.
Christ on crutches!

funny

[Neotrantor]

These guys brag to eachother about how cool their kernals are and trashing ms.. meanwhile my XP box has been running for the better part of a year save for an occaional shutdown to cool off (i use no sides on my case and 2 fans). If these guys wanna agree that windows is a low bar to shoot for these days then their obviously in denial.

possible answer - reliability, stability

[Roadkills-R-Us]

At work we have a compute farm that includes both Solaris and Linux. How many of each we run is based on the software requiements to do our work, of course.

Overall, Linux does a great job. But we experience odd lockups we can't easily track down. The only alternatives seem to be pulling software developers from their real work to debug the kernel, or paying fat licensing fees to one of the Enterprise class Linux vendors. At that point, Linux is suddenly in the same arena as Sun, WRT price. Of course, there's always the option of simply replacing the hardware; it is fairly cheap compare to Sun hardware. Now there's a green thought. 8^/

And for the monkey's edification, some of us do care about library compatibility. I've certainly run into issues.

And for the record, I haven't been able to get my sound card at home to work on Linux ever since I moved into the 2.4 kernel space.

Linux is a good thing. But so is Solaris. And "Use the source, Luke" is the wrong answer for the average end user-- even the average technical end user. It reminds me of why I picked Linux over BSD almost a decade ago. ``Just write your own damned driver and quit whining.''

If I start hearing much more of that, I'll start looking for an alternative to Linux in a heartbeat-- and I'm referring to the compute farm at work as well as this system at home.

Re:What is up with you armchair kernel hackers?

[Anonymous Coward]

>But Linux isn't ready for the big iron machines Solaris dominates yet.

I see your frustration with the slashdot crowd
not reading the article, but I have to point
out that I don't agree with this statement.

What does IBM sell, if it's not big iron running
virtual Linux boxen?

What about SGI's Altix 3000 mainframes with
512 processors using numa (soon to be 1024 processors). If that's not big iron, I don't know
what is.

I agree that Sun knows something about engineering principles, but they are so
obsolete in terms of applications. From
a developer's point they suck. No compilers,
no editors (except for vi --- not vim, just vi),
no real shells (Crappy SHell is a not real shell). Perl? Python? What are those?
How about Java, they're own language?

Can you really think of a suggestion where
Sun has a solution that works, and that's better
than anyone elses? Reliability isn't it ---
consider Google --- talk about fail over
capability and redundant file systems, a nuke probably couldn't knock them off the air.
Speed ain't it either -- they get whooped on
by Linux even on their own hardware.
Price? That's a laugh. What do they have left
then? I'm not trying to troll -- I'm being
serious -- in what REAL WORLD scenario
do they have a better solution? I know that
they still win contracts, but I don't believe
they are winning the contracts on the
merits of their products, more likely
the ignorance of their customers' to alternatives
to Sun.

Solaris is superior to Linux in many ways

[Serveert]

I have used Linux for years but I've also used Solaris. Solaris is simply more reliable and more fault tolerant hardware-wise. It's a fact and as Solaris is opened up and more people become aware of it, it will be obvious. Linux is a great OS and works wonders but it's not up to Solaris standards in many ways. Likewise, Solaris isn't as widely used as linux and doesn't support nearly as many peripherals and isn't as good on the desktop.

That said, Sun's cash cow or former cash cow was its hardware not software. Solaris was a nice OS that was icing on the cake. Now that their cash cow is gone, their emphasis will be on Solaris but there's less revenue here. I hope they go bankrupt and GPL solaris personally. :)

The rebuttal wasn't a rebuttal either. It didn't mention kgdb which allows you to debug kernels using source code.. it can also work with UML kernels. Also the rebuttal didn't address the points raised:

Reliability - Reliability is more than just "we're more stable than Windows." We need to be reliable in the face of hardware failure and service failure. If I get an uncorrectable error on a user process page, predictive self healing can re-start the service without rebooting the machine and without risking memory corruption. Fault Management Architecture can offline CPUs in reponse to hardware errors and retire pages based on the frequency of correctable errors. ZFS provides complete end-to-end checksums, capable of detecting phantom writes and firmware bugs, and automatically repair bad data without affecting the application. The service management facility can ensure that transient application failures do not result in a loss of availability.

Serviceability - When things go wrong (and trust me, they will go wrong), we need to be able to solve the problem in as little time as possible with the lowest cost to the customer and Sun. If the kernel crashes, we get a concise file that customers can send to support without having to reproduce the problem on an instrumented kernel or instruct support how to recreate my production environment. With the fault management architecture, an administrator can walk up to any Solaris machine, type a single command, and see a history of all faulty components in the system, when and how they were repaired, and the severity of the problems. All hardware failures are linked to an online knowledge base with recommended repair procedures and best practices. With ZFS, disks exhibiting questionable data integrity can automatically be removed from storage pools without interruption of normal service to prevent outright failure. Dynamic reconfiguration allows entire CPU boards can be removed from the system without rebooting.

Observability - DTrace allows real-world administrators (not kernel developers) to see exactly what is happening on their system, tracing arbitrary data from user applications and the kernel, aggregating it and coordinating with disjoint events. With kmdb, developers can examine the static state of the kernel, step through kernel functions, and modify kernel memory. Commands like trapstat provide hardware trap statistics, and CPU event counters can be used to gather hardware-assisted profiling data via libcpc.

Resource management - With Solaris resource management, users can control memory and CPU shares, IPC tunables, and a variety of other constraints on a per-process basis. Processes can be grouped into tasks to allow easy management of a class of applications. Zones allow a system to be partitioned and administrated from a central location, dividing the same physical resources amongst OS-like instances. With process rights management, users can be given individual privileges to manage privileged resources without having to have full root access.

And of course windows is but a Play Thing.

Re:I thought we had gotten over this already

[greed]

Solaris on x86 is a joke and nobody would use it unless they have a very special need.

Like Solaris kernel development?

Like trade-show floor Internet kiosks?

Like Lego robot control?

What else did Sun employees talk about doing with a Solaris x86 system at SunNetwork '03.... (Didn't get to go this year.)

Solaris wins in big embedded applications

[isdnip]

I'm not a developer, but I deal with different types of systems, and appreciate both Linux and Solaris for their respective strengths. In the telecom space, for instance, Solaris is well respected for building embedded applications. While AT&T invented Unix, they never meant it for critcal "five nines" real-time telephone call processing. Yet the dial tone on my desk comes from a Solaris-driven central office switch. (Not Lucent!) While the switch vendor's own code has crashed, the Solaris layer beneath takes a lickin' and keeps on tickin'.

I think the big fallacy in Linux is the driver ABI. Linus likes to change it, as a way of forcing hardware developers to have open-source drivers. Nice Stallmanesque politics, but impractical in the real world, for at least two different reasons.

1) Not all drivers can expose the source. This is often because complex devices hide proprietary details in the code. nVidia does that with its "compile in the stub" 3D drivers. Even more limiting are the wireless-card drivers, wherein regulatory approval is dependent on limiting user access to some of the chip registers which, in an open-source driver, could be used to create out-of-band or over-power emission. Life ain't all Ethernet cards nowadays. I had No Fun trying to make a PCI wireless card work with Linux, partially because of the (older) version dependency of the vendor's binary-only driver. Solaris and indeed most (not all) Microsoft OS versions have been better about that.

2) There's a lot of custom hardware out there. Sure, Linux users generally think about "computers" that are either "desktop" or "server" systems. But embedded systems are even more common. Solaris works in a lot of big ones, like aforementioned telephone switch. Some of those systems use different makers' boards; said phone switch, for instance, is made by a company that buys critical boards from other companies. Changes in the ABI would make a difficult revision process even harder. And even if you make your own peripherals, having to recompile or, gag, rewrite the drivers to meet Linux' latest idea of an ABI is, well, a serious pain in the kiester. Very unprofessional!

So while most mainstream dekstops do get better support in Linux, in part because of the better volume of applications, the Solaris approach still wins for those big systems where an hour of downtime is worth tens of thousands of dollars.

Re:Because userland Solaris is dead?

[Anonymous Coward]

UFS journaling: done in Solaris 10 build 68 (may happened earlier, noticed it in this build)

JDS will be fully integrated into Solaris 10...

from sun.com

Re:GNU OpenSolaris

[JonAnderson]

For FUCKS SAKE, I am soooo tired of these posts. GNU tools have been shipping with Solaris since Solaris 8 ( year 2000). If you want OSS then just go and get it the same as you would for a linux distro. Try out pkg-get from blastwave - it's very good. The reason why Sun seems archaic in some areas is backward compatability. You would be surprised how many people still rely on sh etc.

Re:Because userland Solaris is dead?

[ximenes]

I hate Sun's tools as well, which is why I spend tons of time replacing it from NetBSD's pkgsrc or elsewhere.

But they have at least made UFS journaling the default in one of the later releases of Solaris 9.

Re:Linux versus X

[ezzewezza]

And, most importantly: Linux Versus Linux. (No you can't actually read it..)

or can you? [archive.org]

Jon Schwartz

[Pros_n_Cons]

If we are going to post Suns blogs shouldn't we post the Red Hat exec's Blog defending [redhat.com] against Sun?

Re:It's typical

[slipstick]

Wow. Funny but I read the two blogs totally opposite.

The Sun guy says "Linux developers don't see the value of features X,Y, and Z..."

And the Linux guy says, "Sure we see the value, we just haven't had anybody provide a good enough implementation to make the pain worth the value. But for those that feel the possible pain is worth it, the features are supported by A,B and C".

The Sun guy than goes in to how Suns implementation is so much better etc. But of goes this wasn't the premise of his first blog, which was that Linux kernel developers didn't care not that Sun's implementation is better.

Re:What is up with you armchair kernel hackers?

[Rambo]

I always love it when people make generalized comments like "it never ever goes down"; referring to Solaris. I spent several years on a team of sys admins helping to maintain anywhere from 5 to 10 E10K boxes running multiple domains on each system. I can assure you they DID go down, and often enough to really cause issues. Mind you this was a very stressful environment, processing millions of telecom records a day, but we got bitten by a myriad of odd bugs, ranging from the eCache bug to random reboots for no apparent reason. No messages, no logging, just poof! Reboot. Then there were the days spent down because some odd hardware fault would keep the on-site Sun guy scratching his head, be it a bad backplane or some other problem. We constantly had issues with correctable memory errors as well; we were told that a few were fine, and it took 50 an hour to get them to swap RAM out. I never got a chance to run those same loads on Linux so I have no basis for comparison, but I assure you there's nothing flawless about Solaris (we ran 2.6-8, incidentally). Much like another vendor's "Unbreakable" claim...

Self-Healing & Hotswapping || But Patch Reboot

[xcomm]

Solaris may still be really reliable. All this Self-Healing & Hotswapping may be nice, but what me very much is making angry is this:

Nearly 50% of the needed patches need single-user mode to get installed and nearly 75% need a reconfigure reboot after applied.

I never need to reboot a Debian GNU/Linux production system that much to hold it up to date.

PS: And Solaris has to be realeased under the GNU GPL too be really cool!

Re:What is up with you armchair kernel hackers?

[youstupidbigot]

and I've seen Linux carry on with about every hardware failure possible, and when Linux has found one of these hardware failures, it reports it, and keeps running as much as it can.

Yeah, it's great when hardware fails but the OS just spits out some error messages and keeps plugging along. It's espcially great if you like data corruption.

OK, sarcasm aside, my point is just that hardware failures can cause data corruption, and you want your OS to be aware of this.

Solaris knows what type of errors, occuring during which type of operations, can and can't cause data corruption. It does the right thing in the event of hardware failure.

Re:Not much longer

[Kalak]

or my favorites:
Sun v. SCOX [yahoo.com]
RedHat v SCOX [yahoo.com]

Shal we add any other UNIX vendors in there (SCOX was^H^H^His a UNIX vendor afterall.) No other "pure" UNIX vendors come tom mind (HP sells too many PCs, same with IBM, etc.)

We could add Sun v Apple [yahoo.com] since they are both a hardware and an OS vendor (even BSD flavored now). I don't believe Terra Soft is listed, but we could compare there as well.

The 2 year timeline avoids stock splits, as Sun has two earlier ones.

Oh heck, let's go for Sun v. RHAT v. SCOX v. APPL v. MSFT [yahoo.com] for fun.

Re:Why

[caluml]

Today we have frequent kernel panics and X server flakiness, gigantic frameworks for desktop environments and gigabyte sized base installs

I think I speak for many people here when I say: TROLL.

Re:Hardware Support

[slipstick]

Except that your missing that Linux supports PPC, Sun hardware, different and sundry handheld devices and tonnes of other platforms other than i386 based.

So the proper statement is that "Windows supports more hardware out of the box for i386 machines, but Linux supports more hardware in general." The latter part is what people mean when they say "Linux supports the most hardware out of the box than any other OS."

Anecdotal tide turning tales

[Builder]

First off, a short tale about Sun. I recently bought a V20z dual opteron rig from them. On two separate occasions, after logging HARDWARE support calls (faulty ram and faulty powersupply), they've phoned me within 2 days and asked why I'm running Linux on the machine, and have I considered running Solaris instead. On each occasion, I've told them that we have no interest in Solaris on x86, but they've gone on to give me a hard sell.

They may well be a company that supports Linux, but they're pretty damn schizo about it :)

Re:Two Points for Debate

[macsuibhne]

It was a hardware bug; the chips that Sun was using were susceptible to a "couldn't happen" hardware bug, where a bit in the cache would flip for no reason (cosmic rays?). The Sun hardware folks hadn't allowed for this in their design. When they finally figured it out, the fix was _hardware_ error detection/correction in the cache (McNealy: "we engineered it out"). The chip vendor knew about the problem, but didn't bother to tell Sun. Chip vendor's name? I.B.M.

Re:Does Sun love Linux or hate Linux today?

[Splatta]

To address your questions:
- Is Solaris supported on Intel86 architecture or not?
Yes, very much so. Some of Sun's highest selling servers are x86 (but based on AMD): link [sun.com]. It's also rumored (yes, I know, rumor, but wait a few months 'till it's out) that Solaris 10 is currently the fastest OS on x86.

- Does Sun sell Cobalt appliances or not?
Not anymore, those have been EOL'd, and rightly so, that was a bad purchase. link [sun.com].

- Does Sun resell Linux or not? Today, is it RedHat or Suse?
Yes, they resell Linux, and they sell SuSE, Red Hat, the Sun Java Desktop System, and the Java Enterprise System (based on SuSE). link [sun.com].

- Is Java a programming language or is it a more General Product? What does "Sun Java Desktop" have to do with Java?
Well, it would seem that the Java Programming language is a language. But it would also seem that Sun wants to leverage the familiarty of the name Java towards other products (Java Desktop, Java Enterprise System) . Not the best marketing move but thats not what Sun spends most of it's money on.

- Can I redistrute the JDK with my own applications or not? Wait, just javac?
I dont' know.

- Is Java called 'Java', 'Java Two', 'Java one-point-two-and-above' or 'Java Five-point-oh'?
If you're going to bitch about naming conventions, Java would be a great candidate, but I'm sure there are many other deserving software packages.

- Where is Java installed today? /usr/j2se ? /usr/jre1.4.1_05b1? /usr/java? /usr/java1.3? C:\jdk1.4.1_03? C:\Program Files\jdk1.4.1_03??? C:\Program Files\Java\j2re1.4.2_04 ?
Doesn't it ask you where it should install? I think this is your fault as the administrator of your own computer.

Re:Cameras

[spinlocked]

Well then, my SunVideoPlus (really a Sun badged Osprey 15something) works just fine on my Ultra60. I use it *every bit* as much as I do my PC's Logitech USB camera.

Thanks for asking!

I hear you.

[Estanislao Martínez]

You really ought to try FreeBSD. Though I doubt the laptop support will be any better than Linux, but, to tell you the truth, if you want a Unixy laptop, a Mac is the answer. (They're not flawless by any means, but they will give you far less trouble than a Linux one.)