Slashdot Mirror
SGI's Linux Server
More details of SGI's upcoming Linux server have emerged. According to the article, SGI is already shipping a 4 processor machine, and plans to ship 8-way and 2-way machines at a later time. Both Linux and NT are available pre-installed. The servers look like they're not your mother's typical x86-based server, and come with some interesting hardware features that I didn't even know Linux supported (hot swap drives). Am I not well-informed, or is it possible SGI has some patches (that they're hopefully itching to give us)? Regardless, I'm glad to see a big Unix vendor shipping Linux, and touting it so highly.
[Note that SMP is very different from clustering, so forget all you've heard about Beowulf, Mosix, and such for this discussion.]
Yes, Linux supports SMP. If you have a bunch of user processes that are doing mostly user-space computation, then you should get a mostly linear speedup even with a 2.0 kernel. But if the processes depend on a bunch of kernel services, watch the kernel version carefully.
You see, with SMP the kernel has to be sure that two processors don't try to modify the same data structure at the same time so as to avoid confusion (e.g, deadlock or a crash). The 2.0 kernels supported SMP by putting one big lock on the kernel, so only one process can be doing something in the kernel at a time. One of the major features of 2.2 is that this one big lock was broken up, allowing Linux to scale far better. Unfortunately, there are still many places where the locks need to be broken up further, and work is ongoing.
I believe that it was such a locking issue that caused Linux problems in the Mindcraft benchmarks when multiple network cards were in use.
So the performance of SMP will vary depending on your application and your system. (I've heard, for example, that building the kernel on a 4-way box gave a 3.7 times speedup in one case, which is pretty good.)
Note that many of the commercial Unixes scale better than Linux (i.e., have finer grained locking). I know Solaris has a very good reputation for large numbers of processors. I would suspect that Irix must be good, based on the systems that SGI sells. Linux is getting there. Upto 4 CPUs should be fine for most tasks. I haven't heard reports of Linux with more than 4 SMP CPUs.
Cheers,
Joshua.
--jon. Postel is dead. May we all mourn his, and our, loss.
I spoke to an SGI insider about this a while back. His comment (anonymously) was that the issues that hold back an SGI "workstation" (ie., X with accelleration for the Visual Workstation) are mostly legal; apparently, SGI had to sign some agreement with Microsoft in order to get the information they needed to get NT working on the box.
According to this person, (Surprise!) Microsoft insisted on language to the effect that SGI wouldn't support a competing OS environment on the same hardware. SGI's strategy for getting around this is supposed to be to focus their Linux efforts on things that can be branded "servers" until the agreement runs out.
I don't know how true any of this is (I don't have enough contact with the source to establish how accurate their statements are), but it may well be that the technical issues in getting X working right on the Visual Workstation aren't the only problems that SGI faces.
It could be just supply and demand...
Just becouse the managers ordering the computers know what Linux is dosn't mean they know what will happen on a Linux box with 8 Xenon IIIs.
My guess is SGI found a way to make it work but don't count out the "They'll buy it anyway" factor. It's not SGIs style to market to the idiot quotent but you never know.
I don't actually exist.
Cheers,
Joshua.
--jon. Postel is dead. May we all mourn his, and our, loss.
So, I guess, after review, the real questions are:
What are you at liberty to say?
Actually, they don't seem much different in price from the Origin 200 servers running Irix. Why buy Intel when you can get purebred SGI?
D
----
They have, with other manufacturers. It's apparently a standard part of the Windows licensing agreement.
Expect more super-duper Linux announcements from big companies as we get nearer to showtime. Watch with glee as they try to out-do each other in Linux support. It's gonna be an awesome show! Wheeee!
Wonder what they're doing to make 4-proc efficient? 8-proc? Good question. The file system and networking improvements are supposed to allow Linux to scale into that range, but that's only a best guess. I've never seen or heard of anyone demonstrating resonably good >4-way scalability on an intel platform with any operating system. (NUMA machines don't count) Last time I checked, NT couldn't scale past 4procs worth a damn either. What's SGI smoking. Whatever it is, Compaq and a few others must be smoking it too since they're planning 8way xeons as well. 8way K-6 would be a much better value proposition since the EV-6 architecture can scale to 14procs with reasonable performance.
--Shoeboy
Of course, for hot swap to be useful the system needs to be able to deal intelligently with failed drives, generally via RAID. I'm running Linux on a hot-swappable Compaq Proliant 1600 using software RAID-1, which works just fine. I've done the pull-a-running-drive test and watched the RAID driver rebuild the RAID system after adding the drive back in. It's pretty cool.
There is no mention of any nice Linux extras, just that it uses RH6.0.
DWR is Ajax for Java
The SGI machines offering hot swap isn't that big of a deal. Any many linux systems can support it and any that support a RAID controller that supports it, will support it. (Wow, that was a lousy grammatical construction!)
I meant that there are a bunch of RAID controllers that handle the hot-swapping themselves, including the automagic rebuilding of failed drives in a RAID-5 configuration, for example. I think the Mylex controllers are an example of that.
Two years ago I managed to get Linux running on one of HP's high end (at the time) multiproc machines (it had two Pentium 200's I think). For the life of me I can't remember what drive controller was in it -- it was a repurposed NT server, not something purchased new -- but you could pull drives out just fine...
I'd guess SGI's just using a hardware RAID solution, not a software one, which is the only thing that would need real Linux support.
Wonder what they're doing to make 4-proc efficient? 8-proc?
Thought I'd mention, with SCSI drives its also very easy to get Linux to rescan the bus. I have a shell script on my system at home that does that, because I'm a knob and forget to turn on my CDR drive more often than not. Flip it on, run the script and voila! No reboot.
Of course, kiddies, don't try this at home. SCSI isn't really supposed to support powering on a device after the system is powered up, unless it specifically says it can. YMMV.
I'm interested in seeing where FireWire goes from that standpoint. I like the fact that you can disconnect them at will. (Well assuming your OS isn't going to complain!)
- a drive
- a RAID controller
- a SCSI cable
- a SCSI controller on the host
Of course, on many machines you're still prone to motherboard failure at this point, which is why you're probably hooking this up to a Sun E-series box. Make sure the two SCSI controllers that the RAID box is hooked up to are on different I/O boards, so that if one I/O board dies, the other is still there. And of course make sure you have at least two separate CPU/Memory boards, so if one of those dies, the other is available.cjs
The world's most portable OS: http://www.netbsd.org.
Disclaimer: I know little about the Linux kernel. This stuff might all be nonsense.
What if any new functionality has been implemented by SGI as modules, loaded on demand by kerneld? They can be shipped as binaries only without violating the GPL. Since they're not techically part of the kernel, they don't have to be covered by the same license.
Chances are, SGI servers running Linux offer hotswap capabilities thru dedicated hardware. But it's also possible to use software only solutions. There are a few requirements :
:-)
- the hardware must handle it, I mean, you must be able to add/remove a drive to a IDE bus/SCSI chain without everything going mad. Most hardware I met doesn't care if you remove a (umount'ed) drive.
- the software must handle it. With IDE drive, there are ways to force Linux to redetect hard disk geometry (with 2.2 kernels, use modules, with 2.0 kernels, there are unofficial kernel patches to do that. check http://www.enix.org/~skaya/ for an ugly patch allowing to do hotswap with your secondary IDE channel with 2.0.36 kernels)
If you want really good performance/reliability, go for RAID-5. Hardware support is not required anymore. Linux supports software RAID-5 since quite a long time (with the appropriate raidtools), but on-the-fly reconstruction is a recent update. And if you want things like LVM ("oh dear, my 80 gigs pool is full, nah, just add another 18 gigs scsi drive, and poof! I have 18 gigs more free"), you will have to play with latest 2.3.* kernels. It's a domain where Linux hasn't reached (yet) the level of others like HP-UX, but it's improving (it's a big work, because there are filesystems consideration underneath - how do I resize an ext2 filesystem, etc)
A last note about SMP Linux boxen : according to Alan Cox if I remember well, Linux scales very well to 2 CPUs, poorly to 4, and not at all to 8. To solve that, give 'em linux coders octo-xeon servers to play with, I promise they will do their best
Let's settle this silly SMP debate once and for all: get one of these machines to run NT and Linux (or swap out hard drives, I don't care how it's done as long as it's fair...) and test SMP. Preferably with the same applications, but whatever measures of SMP performance that can be done on both NT and Linux... (of course Povray, maybe web serving or whatever apps are sexy this week...)
I know that Linux doesn't have some of the cool features that IRIX does, but it should soundly whip NT, since NT 5 *might* have some of the essential features that UNIX has had forever... (note to dissenters: sure, I'll give you examples if you like...)
pb Reply or e-mail; don't vaguely moderate.
I can't find any pictures with the story. Is there anything on SGI's site about this?
It is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail. - Abraham Maslow
I would need to see some significant proof to believe that this hardware is substantually diffrent than the existing Visual Workstation they sell running Windows NT.
No one has mentioned that they have built a totally new hardware configuration, based on Intel CPU's. Therefore, it's only reasonable to conclude that they are mearly installing Linux on the Visual Workstations, and calling it a "server."
The differance between a server and a workstation is in the eyes of the beholder. If you think just calling it "server" makes it diffrent, I would have to point to the countless users who run httpd, ftpd, nfsd, etc on thier "workstations." But, to me, it's neither, it's just a piece of hardware. What they do with that hardware doesn't change the fact that I believe that the hardware is the same.
So, even if it's the same as a Visual Workstation (which is just a "product name," and doesn't automagically make the product incapable of acting as a server), but even if they took that box, and yanked the video card out, it's still the same hardware, and Linux is know to run on it, and X doesn't run on it yet, so, I just don't see anything new here.
The bus connecting the memory controller and RAM moves data to and from main memory at an astounding 3.2GB per second-six times faster than an AGP 2X graphics bus.-- This would be very good for a server as well.
Combining the most advanced graphics engine available for Windows NT, unheard-of memory bandwidth, and a multiprocessor interface, the memory controller ensures that your most critical data is on the shortest possible path.The tightly coupled Cobalt graphics engine performs lightning- fast 3D geometry, sophisticated shading, lighting, hardware-accelerated texturing, and pixel fill.-- Clearly, they are selling the advantages of thier chipset as "better video preformance." But, when you increas memory I/O rates, and general communication bandwidth in a system, that's going to help out a lot for servers too. Intel Pentium III Xeon Processors Up to four Intel Pentium III Xeon microprocessors provide the pinnacle of Intel processor performance. Your Silicon Graphics 540 workstation features the fastest Intel Pentium III Xeon processors available, with performance-enhancing features such as a dual independent (cache and system) bus TM architecture, dynamic execution, Intel MMX multimedia technology, and a closely coupled Level 2 cache bus running at the full speed of the processor, with cache capacities up to 2MB. -- Clearly, they have a 4 CPU Intel Xeon system, called the "Visual Workstation" that comes very clsoe to thier new Linux/Intel Server claims.
This is all why I suspect that it's the same. Now, there are some features that the Cobalt chipset gives them that will help a server, and it's a lot of work to develop a chipset for a system at this level, so I doubt they will be using something diffrent for the new "server." But at the same time, there are a lot of video and audio subsystems in the Visual Workstation that they could probably remove. And, the focus could be shifted to network capability and away from Video.
So, I guess, after review, the real questions are:
So, coming back at this, I have to think your right, it's a diffrent box, because if they claim "hot swap" they probably have sleds for it. (SGI sleds are WAY to expensive BTW). And, given that one significant differance will probably be in the case, it makes me wonder what the other differances will be... because there may well be many, this might be totally new box, and I sure wish I knew what the specs really were.
Gotta remember to preview... hate it when I forget to put the bold off tag in.. only meant for the first line to be bold :P
In order to sell at even close to the same price as the Windows NT version of their server, they would need to use either a -VERY- cheap Unix (and most of those aren't scalable, high-end, or offer high availability), OR one of the *BSD's, which, for all their merits (which are considerable), don't compete with NT in terms of driver support or software availability.
IMHO, SGI's powers-that-be are no idiots. I believe they picked Linux as the optimal OS to sell alongside NT. That doesn't mean "best" at X, Y or Z, it means optimal, when considering ALL the factors SGI would need to take into consideration.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
This looks like a nice public relations press piece that SGI was happy to get out. Server? Thier Visual Workstation billed as a server? Sure sounds to me like they don't have X working yet.
We knew it would make a good Linux box last year when we saw the memory I/O, and chipset, and all the fancy hardware. Nothing new there, that happened last year. We knew Linux would run on it when kernel 2.2.0 came out and had Visual Workstation patches in it. Nothing new there, that happened last winter.
So, now, we get "news" that SGI has an Intel Linux Server... If that means they are just now getting it installed right, that's sort of slow progress, considering how long ago the first reports of it running were.
Maybe I am just in a bad mood this morning, but, what this story says to me is "Well, we still haven't got X to work, we promised OpenGL, and we don't even have X working yet... We can't get enough people working on it, it's taking way too long, what can we do? I know, let's get some press on it running Linux again, call it a 'server,' and maybe we'll look like we did something good."
Don't get me wrong, I like that SGI is supporting the Linux community. I just don't see any big reason to get excited by this specific story.
First if you unmount a scsi drive with an SCA connector you can pull it out hot, raid or not. This is not something new.
Now my main point I want to make is that the RS/6000s running linux is a much bigger deal. (Yes linux is running on RS/6000s) How does 32 processors sound? Hot swappable processor/memory/drives/cards? RS/6000s will autodetect bad processors and turn them off. When you come into work and your computer tells you processor 5 has been turned off and you replace it, add some ram, and put a new video card in all without ever rebooting your computer that's impressive.
Reasons to reboot a RS/6000:
1. You are moving it to another state
2. All of your processors failed at the same time
3. Your UPS just caught on fire
4. Oh wait, there are no more reasons
Anyone who cannot cope with mathematics is not fully human.