Million Dollar Reviews: Sun E10K/4500/450 Servers 277
redir writes "There is an interesting article on Reviewboard.com about Sun's bigboy E10k million dollar servers. They also have one on the E4500 and the E450.. It's a good read and breaks down the rational behind the architecture designs." I might prefer an IBM S/390 for my own den, but it's interesting for those of us at present lacking a computer budget like these demand to read about what makes them so pricey. Maxing out at 16 multi-CPU system boards and 64 gigs of RAM makes a nice start.
Re:REDIR CAUGHT IN A LIE. *PLEASE READ* Funny! (Score:1)
The E10k frame is capable of holding up to 16 system boards, with a minimum of 4 boards. Each system board can hold four CPUs (480mhz), 4GB RAM (4 banks of 1GB), and either 4 SBUS devices, or 2 PCI devices. (Note: in the future, this *may* support faster CPUs, 2x higher memory density, and 3 PCI devices per system board, specs are as of 03/29/2000)
when i read the article earlier today, the date read 12/29/2000. i wish i could prove it, but i swear it's true. anyone else remember the old date, or have a cached version with the old date? it seems that reviewboard has changed their date to match their "it's been up since march" story. that plus the 480 mhz cpus and the quotes from epinions in the e4500 story (sorry, redir, there's just too much evidence) makes me think that this really is a case of plaigarism. i hope epinions does something about it.
Your Wrong and I'm Right To The Infinite Power (Score:1)
Re:Holy shit, batman (Score:2)
Oh comon, I actually kinda respected u for a sec (Score:1)
THANK YOU FOR THE FINAL PROOF. (Score:3)
BTW, the 480mhz processors for the E10k aren't even available now. See the Sun specification site for details. [sun.com]. Also funny how he knew about Sun's 2x higher memory densitry (which also is not yet supported on the E10k) in March, long before it was available for the lower-end servers.
Re:WHAT THE **HELL**. PLEASE MOD THIS UP. (Score:1)
ARTICLE IS A FRAUD (Score:3)
he E10k frame is capable of holding up to 16 system boards, with a minimum of 4 boards. Each system board can hold four CPUs (480mhz), 4GB RAM (4 banks of 1GB), and either 4 SBUS devices, or 2 PCI devices. (Note: in the future, this *may* support faster CPUs, 2x higher memory density, and 3 PCI devices per system board, specs are as of 03/29/2000)
For starters, 480mhz processors aren't even available NOW for the E10k, and only the E450. See this Sun Specifications for the E10k [sun.com]page for verification.
Now, go read my view at Epinions [epinions.com] about the Sun Enterprise 10000.
At this point, it should be pretty clear which is the original article and which is the changed one. BTW, if the people at Slashdot save a copy of the web pages before they post it, then check their original copy. You'll find that the date was really "12/29/00" at the review site, and not the "3/29/00" that it is now.
It doesn't make sense that a site would say that 480mhz processors are available in March when they aren't even available now. Not to mention the 2x density RAM that wasn't even announced for the lower-end servers at that point.
This story is a plagairism of my original work, and I am COMPLETELY DISGUSTED at the theft of my work.
Re:WHAT THE **HELL**. PLEASE MOD THIS UP. (Score:1)
Selecta
Sun E10k (Score:1)
Re:We have several of ALL of these... (Score:1)
BTW, isn't DSD support for all enterprise multi-CPU getting incorporated into Solaris soon?
Your Working Boy,
Re:hardware can only take you so much (Score:1)
> and tons of disk space, it is DOG SLOW
Want to hire me to set it up properly for you?
Re:No Thank you For being so ignorantly wrong. (Score:1)
Re:How does the community work on these machines? (Score:2)
The S/390 port was started by volunteers which, indeed, did it on their spare time..
At the same time, few engineers from IBM heard about the idea and started to port Linux to S/390 without telling anyone outside IBM and THAT's the port that everyone knows about (it includes a proprietary network driver).
If I'm not mistaken, there is a story about it in Salon's archives.
Re:How does the community work on these machines? (Score:2)
Not sure though...
Re:Oh comon, I actually kinda respected u for a se (Score:1)
here are the facts: parts of two different epinions posts are identical to a review on reviewboards.
parts of another review on reviewboards (by the same author) are identical to yet a third post on epinions.
some people on slashdot claim the editor told them the piece had been up since march and therefore the epinions piece is plagiarism.
a date for specs in the article on reviewboards changed from 12/29/00 to 03/29/00.
12/29/00 is a correct date for the specs in the article on reviewboards. 03/29/00 is an incorrect date for the article on review boards (this is the smoking gun really).
yes, it is possible that there's a psychotic guy on slashdot with multiple accounts, who also has multiple accounts on epinions, who has been plagiarizing (since august - that's when the earliest epinions post about the e10k is dated) bits and pieces of a reviewboards review (that he archived in march, since it was offline in august, remeber redir?) there are people on the internet odd enough to do such a thing.
but it seems much more likely to me that chris chabot's posts on reviewboard are plagiarized from various epinion posts.
as far as redir's, fuckface's, and ataridatacenter's "quotes" from the "editor". whatever. redit has "logs". fuckface was told the article was up in march, redir was told it was up for "almost a year". whatever. totally unprovable in any case, and irrelevant in coming to the pretty clear conclusion that the reviewboard article copied the epinions articles, not the other way around.
and again, the question is, what's your connection to reviewboards, redir?
Re:How does the community work on these machines? (Score:1)
Any problems that you had with the machine would be blamed on Linux, just the time saved from arguing with tech support would be worth running solaris.
Sun has a great os with widespread support, so Solaris isn't going anywhere. Companies like SGI need to dump IRIX because they cannot afford to maintain it anymore, not because they love linux.
Re:Your Wrong and I'm Right To The Infinite Power (Score:1)
Re:More plagiarism from "Chris Chabot" (Score:2)
Um... excuse me, but it seems that you have resorted to insults instead of valid arguments to refute this guy's statements. He has just provided evidence on a third party's site that at least part of the article was completely ripped off. And I'm inclined to believe that the rest was too.
Also, it is awfully strange that someone who just happened to see a cool article on some site would be so zealous in defending the authenticity of the article if they didn't have a vested interest in the article being credible. I sure hope you have good lawyers, because it looks like you're going to be a defendant in a lawsuit pretty soon...
Re:I hate Sun computers. (Score:1)
NOT AT ALL. (Score:2)
I bet, however this review has been on our site for over 2 months. You have
a lot of sand to come to us and say you wrote this. Our review was posted
10/25/2000 at 1:57p.m. We moved it to the front cover of the site for a
promotion we are doing on our server section.
I'd love to meet the "so-called author" of this article. All that E10k hands-on experience and a LOT of details.. And when pressed, the editor says the original date is "conveniently" made two months ago. Riiiiigggghtttt.
The style of writing is my own. The unique concept my own. And I *will* go the extra mile to prove this. This is an INSULT to me.
Re:I hate Sun computers. (Score:4)
Sparc is a lousy processor. 400 megahertz? And software support and development problems are also bad.
That 400 megahertz processor operates on about 4 times more CPU instructions per clock cycle than your X86 chip. You're comparing apples and oranges. And I have bad software support problems on my IBM Aptiva running Windows that crashes every 5-7 days. What problems do SPARC chips have that x86 chips don't?
standard Linux tools like Gimp
What exactly is standard about needing a massive image editing package with your server? Dumb statement
How will we ever be taught about the high level programs the end user deals with or the websites, when we don't even have a graphics tool comparable to Microsoft Paint.
How will you ever get a job in the real world when you equate Microsoft Paint with Oracle in the same sentence. I'm a sys admin and haven't touched a graphics program for work in over 5 years.
Sun computers are expensive, unreliable, slow, of a bad design, and are falling more behind each day.
Expensive? Yep, but they run better than your x86 boxes, even running Linux, sorry. Unreliable? Maybe when you let the developers have root access and tune things to their heart's content. We ran E10 domains that were up for months, and only went down because we installed some new software on a test domain (hence test), or we were installing upgrades and brought it down. Bad design? Maybe if you want it to look like an Intel or Apple, but for what it does, it does it as well as anything out there. And what's with falling behind more every day. Get your head out of your Megahertz, the days when X+25 is faster than X have been gone since the Pentium Pro and a second motherboard chipset.
Amazon is now on Linux.
Because Amazon hasn't made a dollar in over 3 years of operation. They can't afford Sun.
selling, hosting static pages, sharing information, databases
Selling? What's that got to do with computers? E-Commerce you mean? It's all in the software. Of course, if your computers can't handle the load, then you've got a problem. Hosting static pages? Yay, whooee, big load on your computer there. Static pages aren't where the web is going anyway. Sharing information? That's what Email is for. And Databases? On Win2K? Maybe if you're talking about your contact manager database with your friend's names and phone numbers, but for that matter, you could have used a CSV Spreadsheet from Excel for Windows and a little DOS batch file to break it out.
Sun Bigot? No, but kiddies who convinced their parents that they needed to have that 1GHz Athlon because it's 1000KHz and so it's the fastest, and have no concept of system architecture irritate me.
Data warehouse (Score:3)
YOU ARE THE ONE WHO SUBMITTED THE ARTICLE! (Score:2)
redir writes "There is an interesting article on Reviewboard.com about Sun's bigboy E10k million dollar servers.
And you seem to have some really inside knowledge about the site. RIGHT. It appears obvious to me that you work for the site that hosted the article and are now trying to cover some tracks. CONGRATULATIONS. YOU'VE BEEN CAUGHT.
EVEN MORE LAUGHABLE REPLY FROM "THE EDITOR" (Score:3)
The author works for Novadigm and is currently working with over 100 of these machines in an ongoing project for the U.S. Government. He holds a PH.D. in computer science and is a well known individual in good standing. You sir are a fraud.
100s of E10ks? 100s!??!!? **NO** site has 100s of them. The largest customer, as far as I am aware, is AT&T. The second largest company that uses E10ks is the company that I work for, and "dozens" may be stretching it quite a ways.
Not even AT&T has hundreds of these machines. Your latest statement is further evidence of the fraud going on here.
Re:EVEN MORE LAUGHABLE REPLY FROM "THE EDITOR" (Score:2)
I wouldn't be surprised if the NSA had 100's of them. I would be surprised if someone who worked there ever told anyone about it.
Re:WHAT THE **HELL**. PLEASE MOD THIS UP. (Score:2)
Then someone else(*1) should submit the abuse complaint to epinions. One way or another, epinions should take a look at it.
I'm not wasting my time on it though. Well, other than these two posts :)
(*1) Such as yourself or one of the other people who are railing against him.
Re:FOLLOW THE MONEY (Score:2)
Maybe AtariDatacenter is telling the truth (Score:4)
At the moment, I would not categorically say that I fully believe him, but until I hear further or learn more, I would not dismiss AD's claim just yet. Here's why:
Anyway, as I said before, I would not place money either way ... but I wouldn't discount AD story just yet either.
Hey AD, if you are pursuing reviewboard for plagarism, how about updating Slashdot on the results?
Re:our friend's identity (Score:3)
This is a brilliant bit of research, proving that the guy who submitted the article to Slashdot is the same guy who runs the site, although he now is trying to hide the connection. And he's the only guy who's arguing vociferously that the plagiarism claims are bunk. Very curious!
If I could give this poster some of my karma, I would. But I can't. Could a moderator please throw him a bone?
RB article can't be from March; it's not in Ggl (Score:3)
If you go to Google and search on phrases from the ReviewBoard article that are likely to be unique to the article, e.g. "E10k frame is capable of holding", how many hits do you get? Zero. This suggests that the article is too new to have been indexed by Google.
However, redir asserts that he has reliable information [slashdot.org] that the ReviewBoard article was written in way back in March and therefore couldn't have been copied from AD's epinions piece from a few days ago, as AD claims.
If the RB article is actually nine months old, why hasn't Google indexed it yet? Certainly, Google combs RB more frequently than once every nine months. For example, here [reviewboard.com] is a RB article posted in March, and it's in Google: Search on "sprint pcs makes me think" [google.com], taken from the article's opening line. Google returns the article as the first hit. (Following hits come from rb.chabotc.com.)
In summary:
This evidence is highly suggestive that AD is correct and redir is not.
Re:Data warehouse (Score:2)
Guy'd have to trip over BOTH cords to take down my system (but PG&E can still do it quite nicely - no UPS - it's a test system anyway, nuthin mission critical going on here.)
It's a really nice system, but for $20k, you'd think I'd get a frikkin floppy drive. What is this, the biggest iMac ever made? Wait, it's headless, that can't be an iMac.
Re:We have one of these.... (Score:2)
NOT ON A SINGLE SEARCH ENGINE. SURPRISED? (Score:2)
Rectification (Score:3)
As a result, i've imediatly have taken the articles down and we will attempt to contact the user in question and have some harsh words for him.
As far as redir's comments on slashdot go, he's a good friend of mine who's known me for a long time, and felt it was imposible that i would do such a thing, thus tried to defend my name to the bitter end. The real situation wasnt clear to him then either.
New articles on the Sun servers, are being written as we speak, based on the real specs, and based on our own experiances with the E4500 and E10k (i have worked with one such a beast for one of my customers).
My, and our, apologies for the situation, we will try to be more vigilant in the future to avoid such situations. However they can never be fully avoided, even epinions.com reguarly gets faulty and non-unique user submissions. The only action you can take on this is remove the article in question as soon as posible.
Now first some coffee.. its way to early for such a headache..
-- Chris Chabot
"I dont suffer from insanity, i enjoy every minute of it!"
You're getting there (Score:2)
You'll notice that user's site info is given as "www.reviewboard.com" and the user's email is "philip@ferreira.net" which happens to be the same as the admin contact for reviewboard.com.
Come clean. Make real apologies. And we can all move on.
Re:WHAT THE **HELL**. PLEASE MOD THIS UP. (Score:4)
Then what you have to do is very very simple.
Remember when you first joined epinions? Remember the big form they asked you to read and sign? Remember the bits in it authorizing them to take legal action against any site that uses your review without approval? Remember the bits asking you to inform them if you ever discover someone ripping off your review?
Excellent! Go here [epinions.com] and report these violations, then sit back and let epinions' lawyers whip their asses.
Admit it, occasionally lawyers are good for something :)
Re:Oh comon, I actually kinda respected u for a se (Score:2)
From dollardude@cwicweb.com Wed Jan 3 18:21:20 2001r d.com)
Return-Path: (dollardude@cwicweb.com)
Received: from mars.galstar.com (root@smtp.galstar.com [204.251.80.4])
by galaxy.galstar.com (8.9.3/8.9.0) with ESMTP id SAA07653
for (jmccorm@galaxy.galstar.com); Wed, 3 Jan 2001 18:21:20 -0600 (CST)
Received: from mail.easymo.com (host2.cwicweb.com [63.117.39.2] (may be forged))
by mars.galstar.com (8.9.0/8.9.0) with SMTP id SAA15545
for (jmccorm@galaxy.galstar.com); Wed, 3 Jan 2001 18:21:19 -0600 (CST)
Received: from gigahertz.cwicweb.com (216.220.240.34[216.220.240.34])by ECOM2(Ma ilMax 4.2.4.7) with ESMTP id 24729271 for ; Wed, 03 Jan 2001 18:28:39 -0600 CST
Message-Id: (5.0.2.1.2.20010103191124.00acf1f0@mail.reviewboa
X-Sender: dollardude@mailpo.cwicweb.com
X-Mailer: QUALCOMM Windows Eudora Version 5.0.2
Date: Wed, 03 Jan 2001 19:19:50 -0500
To: Josh McCormick (jmccorm@galaxy.galstar.com)
From: Dollar Dude
Subject: Re: MY WORK WAS PLAGIARIZED BY "CHRIS CHABOT"
In-Reply-To: Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"; format=flowed
Status: ORr
I bet, however this review has been on our site for over 2 months. You have a lot of sand to come to us and say you wrote this. Our review was posted 10/25/2000 at 1:57p.m. We moved it to the front cover of the site for a promotion we are doing on our server section.
At 05:57 PM 1/3/01 -0600, you wrote:
>RBMAG:3 A 4C813E-prod2
>
>I'm writing to you because it appears that a review that I wrote about the
>Sun Ultra Enterprise 10000 was completely plagiarized and placed on your
>site. Please visit the following URL:
>
>http://www.epinions.com/enth-review-41-2D0E2BF6-
>
>Now visit the review on your site:
>
>http://www.reviewboard.com/Section/Cover/E10k
>
>Look just a little bit TOO FAMILIAR? I should know. I wrote the thing in
>the middle of the night when I couldn't sleep.
>
>I couldn't be MORE UPSET.
SHOW ME THIS "PHD" AUTHOR (Score:2)
Re:WHAT THE **HELL**. PLEASE MOD THIS UP. (Score:2)
Re:Rectification (Score:2)
But that means ... (Score:3)
Either way, this whole thread has shown two things:
Re:ahha plug plug.. goto his language thing and bu (Score:2)
Plugging my site? This looks like a case of the pot calling the kettle black.
Let's review what you have done, since you say that I have plugged my site - and compare.
Whois info:
Administrative Contact, Technical Contact,
Billing Contact:
Ferreira, Philip (PF2861) philip@GWI.NET
Reviewboard Magazine
913 Elm Street, Suite 500
Manchester, NH 03101
603-625-1564
Google search for "philip ferreira" +slashdot turns up:
Slashdot:3D LCD Screen without Glasses
Redir: But yeah... I was wrong, an apology? No I am not sorry for having an opinion
Redir: Err... wait a minute... poor ataristickuphisass(sorry but I'm working out my anger guy) his article was ganked!!
Re:Rectification??? (Score:2)
-- Chris Chabot
"I dont suffer from insanity, i enjoy every minute of it!"
Re:REDIR CAUGHT IN A LIE. *PLEASE READ* Funny! (Score:2)
Anyone who knows Sun equipment knows that the E10k is the last to receive the latest and greatest. It is always released first for the lower-end servers.
Re:Rectification (Score:2)
anonymous coward [slashdot.org]. [...]
Ooh! If that is really redir (and it looks like it is), that's an excellent catch. Thanks for noticing it!
This is where the web is a little scary. I hope this doesn't permanently hurt either of them. Do I
trust ReviewBoard.com? Not at all. But I am not sure that should equal a lifetime ban on employment for those two.
For me, the hiring decision would have a lot to do on whether or not they came clean in public. I don't mind hiring people who have made mistakes; failure teaches some lessons that success never does. But people who can't admit their errors are dangerous, and likely to repeat them over and over.
If the post you found was really from Phillip Ferreira, I hope he comes back and posts it again with his handle and his proper email address.
Chris Chabot and reviewboard.com still have a lot to answer for, of course. Even if redir's subsequent defense was a understandable youthful mistake, the original articles and Chabot's disingenuous spin control does a lot to suggest that the rot at reviewboard.com goes much deeper than an overdefensive sysadmin.
A lesson for redir/editor (Score:2)
This whole thread really has highlighted how objectiveness, respect, and honesty are essential qualities in life and especially in running a business.
It is admirable that redir will stand up in a public forum and try to defend his friends and he is obviously passionate and cares about his work (based on the intensity of his replies). These are all good qualities. I personally believe that redir/editor made a couple of silly mistakes at the start and from then on slid down a very slippery slope.
But that reinforces the lesson. Once you get on that slope, it's very difficult (and much more embarrassing) to get off. Redir/editor had a chance right from the start to fully disclose his relation with reviewboard.com. He had a chance to give Ataridatacenter respect and investigate AD's claim. If reviewboard.com had pulled the article from the start and maybe provided a link to epinions, I would have had the highest respect for reviewboard since they would have acted ethically.
Unfortunately, we know this is not what happened. And by trying to defend his friend, but not being polite, redir caused a lot of people to be upset with him. Worse and ironically, he has damaged the very reputation of reviewboard.com that he was trying to protect.
I know I probably sound a bit pompous in this post with a 'holier than thou' attitude and I apologise for this. However, situation/mistakes like this one are one of my pet peeves. Why? Because one of my team members was fired for doing something similar to redir. In my case, we support the national telco and his actions caused a major outage. As team leader, I tried to stop him getting him fired, but management and the client kept highlighting how he was deceitful, pig headed, did not confirm facts and was not willing to acknowledge that he had made a mistake (sound familiar to this situation?).
I honestly hope that redir learns from this experience and doesn't end up in a situation like my ex-team member.
Re:- Really From The Editor (Score:2)
Re:I hate Sun computers. (Score:2)
How's the scalability on the E450 past 4 processors? Non-existent you say? You mean I'd have to spend $223,000.00 [sun.com] on a Sun to even begin to think about something beyond quad-processors? I could spend a 1/5 of that for an 8-processor Compaq/Dell/HP and get better tpc-c performance than the E4500.
The million dollar bet was bullshit and anybody with some sense knew it. Sun's best posted tpc-c score for a non-clustered server is ~150,000 for a 64-way E10k running Sybase for a cost of about $7,000,000. Compaq's best submitted performance for a 4-way server is ~35,000 with Win2k and SQL2k for around $500,000. So basically I get 1/4 the performance for 1/14 the price. Suns cluster really well, you say? The tpc-c number one position right now belongs to a clustered Compaq running Win2k and SQL2k.
Despite these numbers, I agree with you, but not on the basis of performance. When I let my own personal preferences and emotions come into play I like unix a whole lot better than I like Windows anything. As a matter of fact I despise Microsoft Server products, but that doesn't mean that they aren't faster and that they don't get good uptime. Most people's concepts of uptime are based on client systems and sub-optimal configurations that tend to go down. Microsoft server products can be optimized to be stable.
The fact remains that Sun is dropping the ball. Their most impressive hardware just isn't that impressive anymore. The USIII is way way late and not really very impressive. It is an incremental improvement over the USII. Fortunately, it does actually get more done per clock cycle than an Intel P3, but only slightly more. Hopefully they can get the clock speed up.
In the end, if I cared about cost, performance, and reliability I'd be more apt to run FreeBSD. If I were forced to run Oracle, I'd definitely consider Linux on Intel. If I needed to scale really well I'd probably use an HP9000 or an RS6000.
I still haven't seen anybody support an argument for why one would use a Sun Box for a particular application over Win2k on Intel. I can certainly think of a few, but that's me.
Re:An E4500 in your own den (Score:2)
Since the operative phrase here is "in your own den", that implies a different level of power reliability being required. The point is that you can do it without hiring an electrician, as opposed to, say, a VAX 11/780, which I've heard requires 3-phase power. You don't even need to rack mount them; the two I worked with came shipped in a standalone configuration which worked quite well sitting on the wooden shipping pallets.
Re:my 2 cents (Score:2)
Keep the hope alive: learn new architectures (Score:2)
Re:Rectification??? (Score:2)
-- Chris Chabot
"I dont suffer from insanity, i enjoy every minute of it!"
How does the community work on these machines? (Score:4)
I would be really interested in knowing. Thank you!
THANK YOU! (Score:2)
Epinions Review [epinions.com] vs Reviewboard Review [reviewboard.com]. The "CIO/CTO" paragraph was a good example.
BTW, has anyone else noticed how redir, the person who originally submitted the article, is so vigirous in defending the authenticity of the reviews? Amazing for someone who just found a really interesting article and then submitted it.
our friend's identity (Score:5)
do a whois on reviewboard.com.
Administrative Contact, Technical Contact, Billing Contact:
Ferreira, Philip (PF2861) philip@GWI.NET
Reviewboard Magazine
913 Elm Street, Suite 500
Manchester, NH 03101
603-625-1564
Then do a google search on Philip Ferreira. Or, better yet, on both Philip Ferreira and slashdot. Seems like our buddy redir used to post to slashdot using both his username and email address...% 22+%2Bslashdot&hl=en&lr=&safe=off [google.com]
http://www.google.com/search?q=%22philip+ferreira
Re:Data warehouse (Score:2)
A "SOFTWARE ENGINEER" KNOWS INTIMATE E10K DETAILS? (Score:2)
You've got a special line to this editor? He doesn't leave me nearly as many eamil as he did you, a random user of his site who happened to submit the articles to Slashdot in the first place.
But, (Score:2)
As I am saying before, I would not say that AD is telling the truth, however, nor would I say that he is lying. There is enough evidence both ways to say that either could be telling the truth.
In which case, one should keep an open mind and see where AD's enquiries lead him.
Re:Software benchmarks?? (Score:2)
If I recall correctly, Oracle and other database companies have something in their EULA that prohibits you from benchmarking their database. Supposedly, the point of this is to prevent Oracle's competitors (which are... who?) from running 'crooked' tests (not optimizing Oracle, or whatever). But it still makes me nervous when a company puts something in the EULA preventing you from benchmarking it...
FOLLOW THE MONEY (Score:2)
REDIR CAUGHT IN A LIE. *PLEASE READ* Funny! (Score:2)
Please come up with another story to explain this. I'm waiting for a good laugh about the 480mhz processors available in March of 2000... especially since I can't even get them for the E10k (only the E450) right now.
Re:If only Windows could do this... (Score:2)
Sure, why not? [unisys.com]
I didn't believe it at first... (Score:2)
... but AtariDatacenter may be right. I'm posting only because I have an established history on Slashdot and absolutely zero connection to AtariDatacenter or redir.
Consider this a neutral third party observation to document something before it can be changed.
From http://www.reviewboard.com/Section/Cover/E4500 [reviewboard.com], paragraph two:
From http://www.epinions.com/enth-review-5999-27A4508B- 39906C31-prod5 [epinions.com], paragraph 1:
The Epinions date is listed as Aug 8 2000. No date was available for the reviewboard article. The Epinions author (nightfall) is not the same author as the disputed E10000 article [epinions.com] (jmccorm).
If only Windows could do this... (Score:2)
Try that with Windows...
REDIR == Owner of Reviewboard Magazine!!?!!! (Score:2)
by redir (philip@ferreira.net) on Saturday April 24, @07:56AM EDT
(User Info) http://www.reviewboard.com
This is too funny. Thanks for the good research! Nice how redir posted some messages with his address for the redir alias as the administrative contact of Reviewboard Magazine! If this doesn't clinch things, what does?
Re:Data warehouse (Score:4)
Two hours? I'm not surprised...the memory check on 64GB would take fricking forever.
Not to mention running fsck on the disk....
(Mandatory for any hardware thread) Imagine a Beowulf of these.....
ObJectBridge [sourceforge.net] (GPL'd Java ODMG) needs volunteers.
We have one of these.... (Score:2)
Are they willing to pay for the extra training AND for experience time before the system goes live? Think at least 5 man-weeks per operator minimum.
Do they realize that you lose MANY of the coolest features (Dynamic re-configuration) with many of the most common setups? (clusters, fiber-channel fabrics)
Do they understand that part of the issue is the storage, and that the amount of storage complexity to fully utilize one of these beasts can be quite large? An E10K will often be set up with Veritas, or sam-fs, or brocade switches. This is a BIG change from a few linux boxes with SCSI raid arrays. Make sure your people can handle it.
They should REALLY wonder if they wouldn't be better off with one of the smaller (450, 4500 etc) SUN boxes instead.
In other words, do you REALLY need this much power, because that much power not only costs in up front dollars, it costs alot to actually keep it running and to take advantage of that much power.
-Anonymous for a reason
Conclusive evidence that AD is right (Score:3)
Doing the same thing for "Chris Chabot", the supposed author of the Review Board article does turn up a couple of semi-sophisticated computer related hits (needs help compiling a kernel [tux.org]), but in these cases Chris has a Review Board email address! ---> chabotc@reviewboard.com
I very much doubt that the review board is running a Sun E10K.
It seems pretty damn certain that the Review Board plagarized these articles. Since "redir" was also the original submitter and a strident attacker of AD, I would give fat odds, that he is "Chris Chabot" or at least a buddy of his.
But you got to hand it to the guy. Redir is a great handle for someone who redirects content from one site to another.
Smoking Gun #23 (Score:4)
One of the points of contention in this discussion has been that someone said they thought it originally said 12/29/00, then changed to 3/29/00. The discrepency between rb.chabotc.com's and reviewboard.com's article is further proof of ReviewBoard's lie and coverup. I hope other people will post verification of what I'm saying before the copy on rb.chabotc.com is changed again. Meta date tag I'm looking at says 2001-01-02 17:58:27. That date is not dynamically updated as another newbie pointed out, as you can see from looking at other reviews on RB.
You'll also find Chris Chabot, the allegedly reported admin of hundreds of Sun 10Ks, to have written reviews of laptops for Reviewboard, and even the article announcing the grand opening of Reviewboard! Sorry Chris, can't change that one, it's archived on Google [google.com].
Chris also used to post on occasion on Slashdot, under user chabotc, and has posted help requests to a linux-kernal mailing list.
Re:Software benchmarks?? (Score:2)
These benchmarks are conducted under a very strict set of rules and are a very good indication of comparitive performance.
The standard test scenarios are a little simplistic so do not be surprised when a system rated at 10000 transactions per minute strugles to do 1000 tpm in the real world!
http://www.tpc.org/new_result/tpcc_perf_results. as p
Also note the total lack of SUN hardware in the top ten!
Re:Software benchmarks?? (Score:2)
SunOS [hostname obliterated] 5.6 Generic_105181-20 sun4u sparc SUNW,Ultra-Enterprise-10000
I work on both HP and Suns, and I must say, they are both nice machines. However, this uptime/hot swap thing is a bunch of crap. The most telling thing was when I started at this job, and wondered why the Sun DBA's hadn't installed the normal startup scripts. Turns out they had some incidents where hardware work was done, and the domain kept rebooting before Oracle could even finish coming up. This is not good for Oracle, to make an award winning understatement. Now, that can happen on any machine, but it had the effect of making those DBA's paranoid about automatic startups. So whenever anything has to be done (like, remember Y2K?), a DBA has to be onsite to manually bring the db up or down. Sheesh. I had a CPU go out on an HP over Christmas, and nobody even noticed since no one was actually doing anything (except the sysadmins noticed, of course).
So it's fine for a data warehouse. But since I'm working on real production systems, I'm waiting for a V class HP (I was hoping for a cluster of N classes, but oh well). I've worked on those before, and they crank.
What good are benchmarks if the puter crashes? For that matter, what good are benchmarks at all? The biggest computer can have users tapping their fingers with some apps.
Re:How does the community work on these machines? (Score:2)
well I'm sure that sun will open up several to the community for development. The same day porcine beings self-aviate and a certain mythological place experiences a cold snap... ;^)
Seriously, this is where the proprietary unix companies make tens or hundreds of thousands on OS licenses for their bigboy hardware. Tell me again why they'd want linux running there for free?
--
And further evidence: (Score:3)
In addition to your statement, the following circumstantial evidence has been mentioned in the mess of posts in this thread. These include:
Seen on a t-shirt around town (Score:2)
SGI Origin 2000 vs. SUN Enterprise 10000 Systems (Score:2)
Sun vs SGI [syr.edu]
Re:Reviews of Sun Ultra Workstations (Score:2)
Bill - aka taniwha
--
Re:How does the community work on these machines? (Score:2)
Picture it in terms of how many people run Linux today in a large environment. They may have a couple of boxes dedicated to running databases, a couple of boxes hooked up to the internet and handing out web pages, a few firewall systems, etc.
Now take that mental picture and move it into one physical box, but with the same independant pieces. Even the "network" between the pieces looks the same TCP/IP stack to TCP/IP stack (just think of "virtual wires" between the "boxes" within the big box).
So you really don't have one Linux system running on this huge piece of iron, but many. Admittedly you will not get the raw performance of the native iron. But for the most part unless you are doing raw number crunching you should not see alot of degradation.
As far as the develper community support, keep writting the tools and applications that use standard interfaces and don't dip down into the machine code level in your code. If the code is written say in C and one doesn't get two tricky with byte orders then there shouldn't be a problem with the code being ported (a simple recompile is all that would be needed).
- Really From The Editor (Score:2)
*final summary of what has happened* (Score:3)
L. Ron McKenzie points out [slashdot.org] that the person who submitted the story (rdir) and defended its authenticity would seem to be the administrative and technical contact of the site which has the review.
There are also the posts (#1 [slashdot.org], #2 [slashdot.org], and #3 [slashdot.org]) which point to the date being changed from 12/29/00 to 3/29/00 in the published article... after the authenticity of the article was questioned. Of course, the story doesn't make any sense at all with the date and the CPU speeds changed.
Dustpuppy correctly points out [slashdot.org] the difference in writing styles between the reviews. A nice non-technical investigation of the situation. Thanks.
And there are numerous posts, such as this [slashdot.org], which point out strange similarites between the reviews on Reviewboard and those written by people on Epinions. And a credibility check [slashdot.org] betweem the alleged author and myself. This recently posted thread [slashdot.org] seems to be pretty damning, too.
I think at this point, the plantiff rests his case. And I'm going to be. 'Night, Slashdot. And thanks for those who were looking for the truth who helped me prove my case. I appreciate it.
But... (Score:2)
We do not reccommend the 450 for use in our Data center as they are physically large. The 420R provides basically similar hardware levels and comes in a nice small rackmount case (4U). As we use Fibre Channel connected EMC frames for storage the lack of internal drive bays is not a problem.
You've answered you own question (Score:4)
Considering the fact that it is very unlikely that there are several hackers (heck, even one) who can afford to buy a $100,000 to $1,000,000 piece of hardware and invalidate the warranty simply to test the viability of porting Linux, I doubt that anyone outside of commercial developers are working on Linux on mainframes.
A quick search on Google [google.com] for "supercomputer" & "linux" pulls up the IBM machines and a bunch of Beowulf style clusters and not much else. Interestingly most of the IBM links are to Los Lobos [zdnet.com], IBM's clustered supercomputer.
Oops, I just did a search for "linux" & "mainframe" [google.com] and found better links which look like they may point to some enthusiast sites after all, such as ROAM [acude.org]. There are also links to Suse's [suse.com] and IBM's [ibm.com] mainframe linux products to be found.
Grabel's Law
Re:We have one of these.... (Score:2)
Re:How does the community work on these machines? (Score:4)
e.g. the S/390 port was mostly done (and is mostly being done, it's quite stable, but not 100% ready for prime time) by IBM, Red Hat, Millennux (a Red Hat partner), and SuSE.
Except for the kernel and gcc, the code base is nearly the same as Linux on other architectures - therefore, having many contributors on this specific arch is not as important as having them on Linux in general.
(Example: Making KDE 2.0 run on S/390 required just 4 lines of changes).
Re:hardware can only take you so much (Score:3)
We run two E450s filling loaded with 4CPUs (300MHz UltraSparc III in one and 480MHz U3 in the other), 4Gig Ram, 100Gig Plus in in multiple drives arranges as RAID 0+1. And they run very fast indeed
Of course for a single threaded application that runs inside L1 cache, a PIII or Athlon box will beat the SUN. But for with multithread or bandwidth constrained tasks the E450s are worths every penny.
I also really like the design of the casing plastic and ironmorgery of the E450s. Built in cabinet for 20 scsi hot swappable Hard drives. 3 Hot swappable power supply boxes. Everything pops apart easier for hardware mainantance. Lovely box.
The downside for the price you can get ten Athlon 1200 1Gig DDR boxes and still have money left over to rack mount then and buy the rack and cabinet.
Re:I hate Sun computers. (Score:2)
Re:Data warehouse (Score:2)
OK>setenv selftest-#megs 1
problem solved. Now all Solaris needs is a background fsck (or jfs) so that part doesn't take forever.
Minor contradiction in the review (Score:2)
A Sun sales rep will tell you that the E10k is a scalable flexible near infinitely configurable enterprise solution. He'd be right on everything except the word enterprise...
Then in the next paragraph, it says:
We rank this product with 3 stars, but only for high-end enterprise situations.
Doh.
Re:Rectification (Score:2)
You also don't account for the fact that during the controversy, the date in the story at reviewboard was changed from [slashdot.org] 12/29/00 to 3/29/00. Did your "user in question" also break into your site and change that, just to make you look bad? You also fail to explain how that even though you only heard about the problem this morning, that someone took down a piece of evidence at http://rb.chabotc.com/Section/Cover/E10k [chabotc.com], a domain that sure appears to be yours.
Sure, it's convenient to explain this as some "user in question". The appearance, though, is that you and your buds got caught plagiarizing and then tried to cover it up. Does anybody really believe that you take user reviews without giving them the tiniest bit of credit and put them under your byline without checking for quality?
The only question in my mind is whether your site is mainly stolen from Epinions or if this was an unusual occurence. You might as well own up and claim it was a one-time mistake; you and your site might escape with at least a little credibility.
The lesson for the rest of us, of course, is that whenever Phillip Ferreira or Chris Chabot go to get a job in the future, a quick search on their names will show this whole sordid tale. Would you hire them? It's an interesting thing to think about...
Not young machines (Score:2)
. .
Both the Sun E 10000 [unixreview.com] (no doubt the attraction of the piece) and the E 4500 [unixreview.com] have been around a while now, as these slightly longer reviews from 1999 remind me. I expect there will have been numerous updates to shipping variations since launch, nonetheless, which I won't check with Sun's docs right now.
Neither yet support the Ultra Sparc 3 [slashdot.org], which is the chip and associated ( potentially) massively (1024) SMP platform [earthweb.com] probably of most interest to anyone evaluating entreprise scale systems right now. Whether Sun have yet fixed the memory / cache problems which apparently still persist, despite numerous fixes, for the USII I can't tell. But if anyone can post a quick summary comparison of cache design between the two chips, and whether there might be a replay of the well publicised memory problems, that'd be darn nifty. US3 has yet to ship in volume with servers, so there may not be any occasional user reports out there for a while.
Personally, I would rather see a story on Ask /. trying to find someone who could write even a short review (particularly of the E 10000) from production environment experience. The story links did not do much for me. I would not be surpised however if Sun has NDAs preventing real world reviews as part of mandatory support contracts for their big iron.
Oh, and for those of you interested in clusters, here's a related snippet [theregister.co.uk] :)
Re:If only Windows could do this... (Score:2)
Now all NASDAQ needs to do is figure out how to *cool* a datacenter. Last time I was in their MD facility it was like 80 degrees. Of course that was a couple of years ago and I'm sure the fixed it by now...
Re:But... (Score:2)
We have a few hundred machines here, and which have the most problems...
Um, well, we have some E6500 machines which have never worked properly... (They hang at random points, Sun are STILL trying to fix them almost three months after they first delivered them)...
Excluding that... We have NEVER had a problem with the memory riser boards on our 420Rs... We occasionally lose CPUs or Memory due to hardware failures, across all machines. We lose disks every now and again (the oldest ones go first) but as they're almost all mirrored this isn't a big problem... Can you say Hot Swappable?
In your case it sounds like something is jarring those memory riser boards loose... I'd check into your environment...
An E4500 in your own den (Score:2)
Well, save up your pennies, because you can run a well loaded E4500 system (including a nice 21" display) from a single 15A 117 VAC wall plug. I've done it. Two E4500's, a 21" monitor, and a 4-disk A1000 disk unit, to be exact.
Although 220 VAC is preferred to give you higher wattage with less amps, it is definitely not required.
Re:To all the /.'ers mentioning "Beowulf"... (Score:2)
The inter-connectivity of local procs (like in a NUMA arch) makes a big difference if they need to talk together. I don't know of a quick way for the typical Beowulf system to communicate inter-proc near as fast. Beowulf's biggest draw back is the network connection (as in your 16 node iMac cluster). This drawback is not as near apparant for tasks that are normally used on NOW (networks of workstations), such as rendering, because the processors do not need to talk together.
Other tasks, such as the odd/even sort (which allows sorting in logN if you have N procs).
So... a Beowulf of these is possible... if you really need all of that power... and it would look different than the typical grouping that you would see these machines in.
Re:How does the community work on these machines? (Score:2)
You're smoking crack. Solaris is only free for <= 8 cpu. After that you pay through the nose. Then add the cost of the support contract, which is mandatory.
--
Starfire is nearly 4 years old though... (Score:2)
People buy hardware for the stupidist reasons (Score:2)
Re:To all the /.'ers mentioning "Beowulf"... (Score:2)
Beowulf is a shared-nothing cluster.
The E10k can be configured as a single-image 64 proc SMP machine, or 16 4 proc SMP machines, or several variations inbetween.
Linux is crap for single-image computing with lots of processors. Even Win2k is better (and win2k datacenter runs on the Unisys ES7000, 32 proc Win2k machine!)
Incidentally, the E10000 is a crap architecture for SMP computing - its basically a big ass backplane for 2-or-4-cpu boards and lcoal ram to plug into. But at the heart of things, its a bus connecting crossbar node cards. Not very scalable, because eventually you just have too much bus contention.. A 64-way crossbar would be practically impossible, and having 64 procs on the same cpu/memory bus doesn't work at all. So the E10k is a hybrid : 2 or 4 proc node cards with local ram using essentially the standard sun4u crossbar. Then each of these boards plugs into the "gigaplane" backplane. Non-local memory requests go out over the backplane.
The SGI O2k and 3800 are done right, comparatively. Thats why SGI can ship a 512proc single-image machine. No one else comes close.
There are problems with the SGI approach though, NUMA can be tricky to tune and some argue that if you've got to tune at all you might as well go straight MPP via MPI or PVM - i.e. beowulf, and get more or less infinite scalability, but little/no support from the OS for shared resources.
WHAT THE **HELL**. PLEASE MOD THIS UP. (Score:5)
All the sudden, this ends up as a review with a different author at another web site? What the HELL is going on? If you have questions, please EMAIL ME. jmccorm@galstar.com [mailto]
This REALLY PISSES ME OFF! MY article pre-dates theirs. Hell, I should know. I wrote it in the middle of the night. And I don't see any date on their publication. I'm assuming it was published today or yesterday. I demand credit for my work. Hell, this is worth an article.
Re:You've answered you own question (Score:2)
I have installed a number of these for Sun customers, and yes, it is normally one person behind the project from the customer's end.
So what do you go through before you buy an E10k, and decide how to configure it?
Well, you're not working on a trivial project, so first you decide how to configure your machine, and THEN you decide on the platform which can provide to your needs.
So you would look at the application first, then the hardware.
You don't spend this kind of money for the sake of it; your project has to warrant it. So you decide what you need, how important it is, and then you can start talking megabucks if the service the machine will be providing is important enough (and your company has enough dough!)
Since it's established that the project is so important, it must be constantly available, fully supported, not go wrong in the first place, but be able to cope with everything from a disk failure to your power supplier's plant going tits-up without losing the service (remember, the data service here is critical, or you'd not be talking in this megabuck league in the first place).
If I'm going to buy an E10k, plus, let's say, an Oracle database, plus a Terabyte or two storage to go with it, plus the backup solution, plus the disaster-recovery solution, then the cost of support and software is getting trivial. Don't bother me with figures under $50,000. That's administrivia.
So, at this stage, of having decided on Oracle, Sun E10k, a few Terabytes of storage, L700 backups, off-site disaster recovery solution, and quite possibly getting two or more of these to cluster them, someone suggests I think hard about which OS to choose.
On the one hand, I've got my Sun support, with its 2hr callout for anything, 24/7, wherever I am (pretty much), damn' fine OS, and a Sun project manager looking after me who could quite possibly get the sack if things go badly wrong for me.
On the other hand, I've got Linux, which is another great OS, gives me load-sharing cluster, but not highly-available databases; and I can get a support contract from various companies, but none of whom have the power to give kernel updates in the case of a major failure, none of whom I could sue for the millions of $$$ I could lose by the project going bad before or even after signoff, but instead, I can trust loads of people who aren't bothered about my company, but are bothered about their OS looking good, to give me support.
Oh, but none of these people helping me now with my OS (and therefore with my database SW too), can get their hands on an E10k, certainly not on one built just like mine, to test things out before giving them to me.
Remember, that I have already spent millions of $$$
It just doesn't seem worth it for Linux to aim at this kind of market without all the infrastructure already in place.
Don't get me wrong; I *love* Linux for low-end machines (relative to E10kBecause you need access to huge hardware as test boxes, hordes of people with experience in building them, from a hardware and software point of view, and the proof that you've done it lots of times before, and can guarantee to take it all away at no extra cost if things don't work out, even Microsoft can't get into the datacenter; what chance Linux?
Steve.
LAUGH: Unknown amg receives prcessors 1yr ahead! (Score:2)
Of course, I'm wondering why the editor got it, and not the author of the article. Further, I'm wondering why the author of the article says they're available "now", when he wrote it in March of 2000.
How many more holes do I have to blow through this story before you give up?