Linked by Thom Holwerda on Fri 14th Nov 2008 21:38 UTC, submitted by pantheraleo
Oracle and SUN The world hasn't been kind to Sun for quite a while now, but with the economic downturn, things are getting worse. Sun announced today that it will be laying off 18% of its workforce, or about 6000 people. In addition, it was announced that Sun's software chief Rich Green has resigned for reasons that were not stated, although as part of Sun's reorganization and cost cutting efforts, many departments are being merged, and the software division is being restructured and reorganized.
Thread beginning with comment 337332
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE: Sun performances...
by Kebabbert on Sat 15th Nov 2008 11:56 UTC in reply to "Sun performances..."
Kebabbert
Member since:
2007-07-27

Yes, Niagara cpus are more expensive. But I think you should factor in price/performance instead of only price.

One Niagara box equals 10-20 ordinary x86 at some workloads, but it is not 10 times more expensive?



http://searchenterpriselinux.techtarget.com/news/article/0,289142,s...

"On a 64-bit AMD processor and Fedora, we could process approximately 200 matches per second of RSS," Whitehead said. "With Solaris 10 on the T1000, this match rate jumped to 10,000 per second."




http://blogs.sun.com/bmseer/entry/mysql_consolidation_on_sun_sparc

Migrated 251 Dell 2950 Linux MySQL servers onto 24 Solaris Niagara boxes.




Even if 10 times as many x86 servers were cheaper, they require more administration, power, repairs and that cost money too.

I think one of the problems is that SUN can not sell to many of these boxes to a customer, they are too powerful for multi threaded work.

STRATA (europe's second largest web host) that handles 1 billion email/day migrated their whole back end to one Niagara T5440 box.



In SAP official benchmark, one T5440 using 4 Niagara cpus at 1.4GHz achieves 14,000points. IBM's "superfast" new Power6 cpu at 4.7GHz achieves 7,000points. But IBM used 3 servers, in total 12 cpus. Compare that against 4 Niagara 1.4GHz on one machine, running virtual Solaris instances.



No, there are not many customers that need two of these T5440 monsters. One of these can consolidate many servers. And they are far cheaper than IBM, for instance.

SUN is the only big dinosaur that has opened up all their high technology. Microsoft has not done that. Nor Apple. IBM, SAP, Oracle. etc. If SUN bankrupts, no one else will dare to open up their entire portfolio as SUN has. How clever is it to give away everything, instead of selling it? If SUN is successful, other large companies will consider opening up all their portfolio.

Reply Parent Score: 11

RE[2]: Sun performances...
by Laurence on Sat 15th Nov 2008 14:31 in reply to "RE: Sun performances..."
Laurence Member since:
2007-03-26


SUN is the only big dinosaur that has opened up all their high technology. Microsoft has not done that. Nor Apple. IBM, SAP, Oracle. etc. If SUN bankrupts, no one else will dare to open up their entire portfolio as SUN has. How clever is it to give away everything, instead of selling it? If SUN is successful, other large companies will consider opening up all their portfolio.


SUN are in a different, more awkward position to the other dinosaurs though.
On the desktop end (Apple, MS) few people really care about open source. I mean sure you get *nix desktop users who sware by FOSS, but they acount for a vastly insignificant number of overall desktop users.

On the server market, SUN was at risk of loosing out to Linux in much the same way as some of the other UNIXs had. Their only way out (as I could see) was by:
1/ releasing some technologies that empowers it's existing products (ZFS being one killer feature for Solaris imo),
2/ and open sourcing to, hopefully, attract new users (sometimes the strengh of a community alone can help sell a product) and hope they stay on board to develop for SUNs platforms as well as the greater potential of them purchacing licences / support for the non-free services further down the line than if said users never migrated.

In my opinion, SUN have played their hand well (I for one have been swung from running Linux and FreeBSD to Nexenta (an OpenSolaris distribution) as my home file and webserver and I'm very happy with the results of the migration thus far.

I do wish SUN all the best - they're one of the few dinosaurs I still route for.

Edited 2008-11-15 14:38 UTC

Reply Parent Score: 4

RE[2]: Sun performances...
by segedunum on Mon 17th Nov 2008 14:44 in reply to "RE: Sun performances..."
segedunum Member since:
2005-07-06

This is exactly what Sun's problem has always been - trying to play King Canute and get the tide to go back. You can't pull out a few niche benchmarks and say that a Niagara machine is worth twenty AMD or Intel ones, because everyone knows that isn't true. They still believe that in this time of fast,, cheaper, commodity hardware that costs a few thousands people are still going to pay tens or hundreds of thousands for the Sun and 'enterprise' badges.

One or two things don't add up to me in those articles either. You've got a company that enjoys running Fedora in production, but is then willing to spend a spectacular sum of money on a Niagara hardware platform, Solaris 10 and a new administrator? Please........... Migrating 251 x86 MySQL servers on to 24 Niagara servers is also the most far-fetched and naive thing I have ever seen. No specifics of benchmarks either. This stuff is not helping Sun.

Sun quietly knew many years ago that they were never going to compete with the x86 hardware companies for raw performance, so they backed themselves into ever smaller niches with SPARC. Linux on x86 destroyed SGI's Unix workstation market and wiped Sun's out as well. IBM has backed Power into a very high-end and lucrative mainframe niche, however long that lasts, but Sun don't have that option. For those benchmarks to mean absolutely anything to you, whatever benchmarks they have actually used, then you need to be running an awful lot of totally unrelated parallel tasks and threads, and those tasks need to be very lightweight to complete in a reasonable amount of time. Almost nobody has those kinds of requirements. People want to do related tasks twice as big as last year in half the time, and neither Niagara or Sun can help anyone there.

Additionally, given the parallel march of Intel and AMD in putting ever more cores into their processors, you seriously have to question the longevity of Niagara even in the niche market it has tried to carve out.

The solution is brutally blunt, simple and honest: Either Sun gets some ROI from its expensive research and development people to get Niagara and SPARC platforms on a par performance-wise with comparative x86 Intel and AMD platforms, or it gets shot of it. Being protectionist of the SPARC platform will only have one outcome otherwise.

Reply Parent Score: 1

RE[3]: Sun performances...
by Kebabbert on Mon 17th Nov 2008 17:20 in reply to "RE[2]: Sun performances..."
Kebabbert Member since:
2007-07-27

"You can't pull out a few niche benchmarks and say that a Niagara machine is worth twenty AMD or Intel ones, because everyone knows that isn't true."

But for some multi threaded work loads the Niagara IS many times faster than an x86. You dont trust the benchmarks? They are fake? Or do you think I claim that the Niagara are in general 20 times faster than an x86 cpu? Where do I claim that? Either way, you are wrong. For some work loads the Niagaras are way faster, and neither do I claim Niagaras are that fast in general. Learn to read. When IBM announce an benchmark, it is not meant to be interpreted as a general statement, right? It is a claim regarding that particular set up for that very benchmark. If you think otherwise, you think wrong. Again. I suggest you study some higher mathematics. That will sharpen your thinking skills. As of now, you are drawing wrong conclusions. SUN has always been very clear that Niagara cpus are for throughput and not single threaded work. If you think that SUN generalizes that statement to mean that Niagaras are 20 times faster in general, then you are drawing a false conclusion. Again. Like when you again and again always state that ZFS requires several GB to just run - when I and other have told you several times that ZFS doesnt. Ive told you many times Ive run ZFS on 32 bit pentium 4 with 1 GB RAM for over a year. Wrong again. We dont reach you, that's obvious.




And, yes Niagara hardware is expensive in relation to x86. But in comparison to IBM and HP-UX, they are at a fraction of the price. And way faster on some tasks. For instance one SUN machine with 4 Niagara cpus @ 1.4GHz is twice as fast as 3 IBM Power6 machines @ 4.7 GHz with 12 Power6 CPUs on Siebel benchmarks. SUN: 4 x 1.4GHz = 5.6GHz is twice as fast as 12 x 4.7GHz = 56.4 GHz. And, a lot cheaper also. Maybe it is just me, but for multi threaded work, I would surely examine how SUN Niagara boxes does for my work load. So what if it isnt fast on all workloads? IBM is not fast on all work loads either e.g. multi threaded work load. You try to say that you shouldnt even consider what Niagara boxes can do for your particular work load, because of some strange reason?

Understand this; Niagara boxes compete with IBM AIX Unix and HP-UX unix, not with x86. Everything you say about the Niagara, also applies to IBM Power6. Or IBM mainframes which are 1 IBM mainframe Mips == 4 GHz x86. IBM Mainframe CPUs are really slow. You can emulate one Mainframe on a laptop with a program called Hercules. In fact, Ive heard of a customer that migrated an entire mainframe to a Solaris box with 4 Niagara cpus, with plenty of power left.




The customer that migrated 251 Linux x86 Dell 2950 servers onto 24 Niagara boxes, did a naive thing? Ah, you mean they didnt test anything first? Maybe it is you who are naive if you think companies change their important infra structure without testing and doing an analysis first?

For instance, STRATA second largest web company in Europe which handles up to 1 billion email a day, migrated their whole back end to one Niagara box with 4 cpus with plenty of capacity left. You are naive if you think they would migrate without testing first.

Frankly, you would get fired if you worked at my company. For incompetence. If you have a solution that are cheaper and 10 times faster and suits your work load perfect, and still you wont even consider it, then you are incompetent. You make the company loose money it could have saved.

I dont understand your frankly, ignorant, comments. "Who needs detection of silent corruption as ZFS offers?" "You shouldnt even consider Niagara because they are not that fast in general". etc. Really weird remarks. Unless, you work for IBM or some similar company. Then your comments make sense.



And yes, the ordinary SPARC are not that fast. That is true, but SUN dont claim that either. The advantage of SPARC is that x86 is a buggy architecture, and SPARC is not. For instance Ive heard that the Fujitsi SPARC64 allows you to back an instruction if it turns out to be bad. Only Mainframes had that functionality earlier.

Edited 2008-11-17 17:24 UTC

Reply Parent Score: 2