Slashdot Mirror
FCW compares Unix workstations
EngrBohn writes "Federal Computer Weekly evaluated Unix workstations by Compaq, HP, IBM, and Sun -- they specified minimum hardware requirements and a maximum price; beyond that, all was fair. They did not include *BSD, Linux, or WinNT due to space limitations. Here's a chart (in PDF) comparing the workstations. IBM's RS6000 43P Model 260 won on technical merit, but it exceeded the $15K price cap. "
Right, they left out SGI, which happens to be a significant player in the gov field. Why is it that SGI is bypassed when people talk commercial Unix? This still mystifies me.
Worse, the evaluation placed high emphasis on admin tools. Around here, with Solaris, Irix, HP-UX, Digital Unix and Linux in one shop, noone is ever going to use SMIT or SAM or whatever. Just too much pain to learn them all.
Unifying sysadmin tasks is best done with scripting, at which Unix excels (as opposed to NT, say).
If your shop has one flavor of Unix only and has few, if any experienced sysadmins, then admin tools are for you. However, I don't think that most shops in the gov domain are small and hence emphasis on administration ease seem misguided.
What equally mystifies me is that the Compaq unit beats the competition 2:1 on CPU benchmarks, but only gets 3 stars on performance. Yes, it was weak in the graphics dept., but how can you loose so badly that you give up a 2:1 advantage? In my workstation I'd rather tolerate weakish graphics than a slow CPU. Of course, we are never told actual performance numbers and how they were weighted. Too bad.
This is a comparison of unix workstations. A PC running Linux is not the same as an SGI running IRIX. Even an SGI running Linux isn't the same as an SGI running IRIX because if you can't run ProEngineer on it, what's the point?
While Linux and *BSD will run some workstation hardware, it is not the native OS on any true UNIX workstation. Besides, just because the operating system is free does not necessarily make the system cheaper. The operating system is a fairly small part of the cost of a workstation, and if it makes the system more reliable, easier to configure, or easier to get vendor support, then it more than makes up for its own price.
Don't get me wrong, I love linux and use it on most of my computers, but the Compaq Alpha with Tru64 UNIX (why'd they have to change the name anyway) on my desk is a much better CAD station than anything running Linux.
-Alison
Wouldn't it be nifty if all the commercial unix companies got together with all the Linux companies and put together a plan for making Linuxconf into a more generic "Unixconf" tool.
One administration interface for every Unix. A dream likely never to happen...
If you want to help others, please speak about what you understand; don't just fly off the handle and shout "AIX sucks" without knowing what you're talking about. It's these uninformed rants and prejudices that are making the Open Source movement look like a bunch of crybabies.
"Those who would give up essential liberty for temporary safety deserve neither liberty nor safety" - Benjamin Franklin,
This test is perfectly fair. The point was not to compare Unix versions, but exactly what you said, which 15,000 box was best. Doesn't matter if it is a Power3 running AIX or a SPARC running Solaris, if it is easier to administer it is easier to administer, if it runs the benchmark application faster, then it is faster. No matter what the Workstation, the common ground is that they compete in the under 15K field, and their worth in that catagory can be measuered.
A deep unwavering belief is a sure sign you're missing something...
My bad.
I read the internet for the articles.
We had to upgrade an (forgot HP model name) from 10.10 to 10.20. We used the "Easy, Really", "Painless", Ignite/UX tool to perform the upgrade. The upgrade failed, and totally hosed all the filesystems. Then we discovered that nobody had ever done a backup on the system because "it wasn't in production yet."
That's my most painful HP/UX memory, although I can't say that my experiences have been as bad as yours!
Of all the comments I've ever posted, this is definately one of them
So the IBM workstation broke the price cap, didn't meet the minimum requirements for disk space, and needed two CPUs to match the performance of the Sun with only a single CPU, yet it wins the comparison?
Thank you for your input. It is always nice to be able to talk to bthe people who did whatever. This is certainly one of the benefits of open source, after all -- accessibility!
I'm not sure about PA-Risc, but, i am sure that Microsoft has allready dropped NT/PPC, and i think they also dropped NT/MIPS. Last time i checked, you had a choice of NT/x86, NT/Alpha, or the upcoming NT/Merced. btw, where did compaq come in? i don't have PDF on this PC...
Opinionated Law Student Strikes Again!
- A.P.
--
"One World, One Web, One Program" - Microsoft Promotional Ad
"Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
Yes, you have a good point. I was looking at it more from a benchmark standpoint. So, like you said, any comparison would be useless. For $15,000 they have picked the best system for the money.
It's not that AIX is unstable, or performs badly,
it's that the ODM and the commands to manipulate
it are a pain in the ass. When I want to do the
same thing on all nodes of an SP, I don't want to
have to use SMIT on every last one of them.
Once you get past that, though, it's just fine.
Would you shut up already?
It seems like the Performance catagory emphasized graphics performance over computational performance. Why does the XP1000 get such a low score? It's SpecFP rating is nearly double that of the other processors (note that the IBM is a dual processor box).
As far as documentation goes, the man pages that come with AIX are terrible while those that come with Digital Unix are excellent.
It seems to be that the IBM and HP should have been hit a little more on their prices. The points lost in the price catagory were more than made up for on the features catagory.
I don't understand the statement about linux in the IBM administration box. What's linux got to do with this machine?
Seems like the reviewer(s) prefer GUI system admin apps over the tried-and-true command line utilities.
The NT4 CD says PPC, MIPS, Alpha, x86, Pentium, Pentium Pro. Yeah, x86 is there three (or maybe just two) times. I think one of the reasons they dropped NT/MIPS may have been because most post-R4?00s don't do little-endian.
- A.P.
--
"One World, One Web, One Program" - Microsoft Promotional Ad
"Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
It also had the best administration, tied for best performance, best install, support and documentation.
As for the 2 CPU IBM box vs 1 CPU sun box, the IBM box had 2 200Mhz CPU's and the sun box had 1 450Mhz CPU so they compare well. If the box is being used for more than one thing at a time, I'd take the IBM SMP setup *ANY DAY* over the single CPU Sun box.
I work in the federal gov't and subscribe to FCW.
Believe me, FCW and the federal gov't in general
live in Internet time minus two years when it
comes to technology. It's "NT this and NT that,"
"You gotta have Office 2000 NOW!", and the usual
tripe.
- A.P.
--
"One World, One Web, One Program" - Microsoft Promotional Ad
"Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
This was a _very_ sloppy evaluation.
:-)
Compaq's XP1000 was the fastest (CPU wise), cheapest and best supported (30 days money back + 24h/7 days a week) workstation tested. For 5k$ separating it from the "winner" (RS/6000 43P Model 260) you may get a lot of extras, top-notch graphic card included. One wonders, how is it possible that XP1000 took the last place in this test?
Unless one states that this is a _graphic workstations_ performance test, the value of describing integrer, floating point, and graphic performance together equals zero. Even in that case, you have to give some numbers.
This whole "test" looks rather odd to me.
bourbaki40@hotmail.com
a happy user of an Alpha 500au workstation
I agree that AIX is stable, but my main complaint is that it does things in weird/nonstandard ways. I can login to pretty well any Linux, *BSD, Solaris, or Digital Unix box and know that the basic admin commands and how they work, with few exceptions, are more or less the same. AIX and HPUX seem completely different.
Also the fact that you can't, (or at least it wouldn't work for me) do things like add a user by adding them to /etc/passwd and /etc/shadow. You have to use their tool (SMIT).
Oh yeah, and I saw a few weird errors and inconsistencies that had no apparent explaination, and hours talking to IBM's tech support (at least their tech support is good :) to find the solution. Like a machine that "forgot" its partition table. Not fun to restore.
"...Is this world not a call I can screen out" --
I never made the move from 9, my 715/64 is ...).
running the same OS it did the day we were
given it in 95(?). It gets decommissioned
next week, for a white box running Linux
and/or FreeBSD (still deciding
I love that excuse. What, did they run out of paper and ink? Warehouse space? Or space in the writers/editors mind?
"Those who would give up essential liberty for temporary safety deserve neither liberty nor safety" - Benjamin Franklin,
The RS6000 may have won, but AIX makes HP/UX look good! And their GUI, unless it has improved dramatically since I used it 2 years ago, leaves much to be desired.
"...Is this world not a call I can screen out" --
They didn't include SGI either, which is strange considering they used a GIS application as their one and only benchmark. Why would you not benchmark a variety of applications which would be more relevant to potential buyers?
That was the least detailed, least comparitive comparison I have ever seen. Where are the tables on system preformance? Where is the list of features and details about what each system has for that price? Where is the valid reason why some systems didn't win?
Oh, they said IBM, I better go get one... (sarcasm). Not that there is anything wrong with IBM systems, but they didn't even specify which one and which configuration was thier winner, and what makes it better. IBM's RS/6000 43P Model 260 according to what they are saying is just under $15,000 and bearly nudges out the competition. What about this makes it hard for me to believe.
If I send them a 1G drive, will they fill it with a detailed comparison and get it on the web, since "space" was the problem in reporting the comparison?
"None of the boxes in this comparison uses Linux tools, so the System Management Interface Tool used by the IBM model is the big administration winner."
It sounds as if they are using Linux as a baseline to compare the administrative abilities of these systems... nifty.
Anyone else notice that the IBM system let's you play Abuse while it installs? Even all you irrational AIX-haters have to admit, that is a step up from Tetris.
Only 4 systems on thier one table, and it's a pdf table... I am a little dissapointed. I suppose buying $15,000 systems isn't the easiest thing to do for a comparison, but I would have thought an SGI would have made the cut of choices. Notice, HP, IBM and Sun only have 9-5 support, wheras Compaq is 24-7, and Compaq had the only money back guarentee if you didn't felt the system lived up to it's claims after it arrived. But that only gave it 20 points more out of the total of 1000 points. "None of the boxes in this comparison uses Linux tools, so the System Management Interface Tool used by the IBM model is the big administration winner." So, why not be fair and add a VAResearch box to the list? I'm sure they could have provided a $15,000 box for testing.
Also, note that the IBM was $17,587 and the Compaq was only $12,514, but that just factored into the score, and they didn't really cap the price at $15,000...
This seems to be a really short list of workstations, not only did they leave out the Free Unix crowd, they also left out the SGI workstations. They also forgot to measure the useablility of these workstations (speed isn't everything in the workstation market) since some of them come with horrible versions of Unix with terrible UIs.
I read the internet for the articles.
This test seems odd to me. Comparing different hardware and OSes at the same time does not seem to be a fair test. All this test goes to show is which $15,000 total package is the best. All versions of Unix are not created equal.
In the article, they asked for the following systems: "512M of RAM, a 9G hard disk, 24-bit video and a 20-inch monitor. This is the minimum configuration, and vendors were free to add on as they saw fit." How is this a fair test?
What they should have done is restrict the test to specific hardware requirements, not a floating scale of hardware. Throwing in different flavors of Unix just confuses the test more. It's like comparing different brands of apples and oranges all together.
The results of this test will only be useful to managers who have no technical backgrounds.
That really makes me feel for who ever maintains the HP-UX systems I use (mostly 715s running 10.20). I haven't had any problems using them for writing papers and programming.
Why ?
AIX is stable. I manage approx 120 standalone AIX-system. None of them has ever crashed (in 4+ years) from bugs in the OS.
There is always some with differant opinions or experiences, but in my mind most AIX installations has better stability than most Linuxinstallations.
Sorry, your post is at best uninformative and worst a silly attempt at trolling/FUD/advocacy.
For a comparison on Unix workstations, they sure managed to keep bringing up how some of these systems can run NT. What is more useless than NT on Alpha anyway?
Of all the comments I've ever posted, this is definately one of them
What is more useless than NT on Alpha anyway?
NT on PA-Risc, NT on PowerPC and NT on MIPS are good contenders. (yes, they do exist) The most useless of all though, has to be NT on x86 - nothing is more useless that this.
--Shoeboy
Reviews are written by a person.
People have biases.
Stats lie.
As with any review, you have to read the whole thing, not just the bottom line numbers. It looks like they had it in for the Compaq and fudged the numbers to make sure it came out lowest. If you read all of text in the pdf chart and draw your own conclusions, you will be making a much more informed decision. If you read different reviews, you will make an even better decision.
You can come up with a set of benchmarks to make a Mazda Miata look better than a Ferrari F355. IT Managers are not stupid (most of the time, anyway). They look at more than one review, they look up *all* the info. It makes no sense to get your feathers all ruffled over one obviously incorrect review.
This may be a lot better than the MindCraft benchmark (please don't start a flame war over this), but it's the same kind of useless complaining. What it really comes down to when comparing UNIX workstations/OS's is what do *you* want to use it for.
-Alison (sorta bored at work if you can't tell)
I've had extensive experience with ArcView. For those of you who unaware it is a "desktop" (read, limited features) geographic information system produced by ESRI. Typical uses include producing maps and the analysis of spatial data. It can be quite graphics intensive as many millions of lines, points etc may be symbolised and rendered to compose a map. It can be reasonably compute intensive if latitudes and longitudes may need to be "projected". Disk throughput is also quite important as geographic datasets can be in the order of hundreds of megabytes.
There are many ways you could use ArcView and this would emphasise the relative merits of different computing architectures. It is not specified in the "benchmark" how ArcView was used.
However, the ArcView UNIX workstation market is VERY limited and I wouldn't have thought there would be as many users as, say, CAD or engineering. Thus, it would seem an inappropriate application choice for a general purpose UNIX workstation benchmark. ArcView is only available on those particular UNIX environments (and NT too) which is another reason why they didn't try to benchmark Linux, FreeBSD etc. By far, the majority of ArcView licences are for x86 NT boxes (I used to work for ESRI Australia, btw).
On a scale of 10, I'd rate the review as "pissweak" - lacking in technical merit, depth and objectivity.
-t.
What kind of comparison is this if it leaves out *BSD, Linux and WinNT? Those are the three most popular operating systems! Plus, *BSD and Linux are free, which should allow a MUCH cheaper system...
-"One machine can do the work of fifty ordinary men. No machine can do the work of one extraordinary man." -EH