Oracle rises for Unix server push Oracle is taking the fight to Unix market leader IBM with its eight-core SPARC T4 processor and systems with rack, blade, and clustered systems – a full data center press. The SPARC T4 processors, with an S3 core, were developed under the code-name "Yosemite Falls" and offer better performance than Oracle expected. They will …
I don't trust Ellison. Never have, never will. He's too unpredictable. Don't believe me? Look at his marriage record ... To me, Sun Microsystems is gone forever; I'll never spec Sun kit again ... Unless Oracle spins it off as an independent company, and soon, that is.
Larry doesn't like me, either. I can live with that ;-)
I cannot see anyone buying a T system that does not already have one and is not ready to migrate.
This announcement was pretty lame compared to Westmere and Power7
3.0GHz with OOO execution (Oracle is acting like out of order execution is a great new technology)
The T chips are about the only chips that don't have it. Even the old USIII's had it
Dedicated L2 128KB cache - wow that is almost big
Shared 4MB L3 cache - 4MB/8 cores has anybody done the math on Westmere or Power7 they should be ashamed of trying to be proud of this
8 cores with Private L2 Cache - lets repeat the prior statement to make it look important and just dumped down from 16 cores
"up to 5X the single threaded performance" I am guessing that is compared to T1? T2? can't possibly be compared vs. Xeon or Power cause I am betting it is 1/5th
The sparc Supercluster idea is pure marketing BS Peoplesoft and JD Edwards performance improved up to 8X? I didn't even think anyone had JDE on SPARC. I thought they killed that before Oracle bought Sun.
<1% virtualization overhead. Well when you pin the partitions to threads which are only on one CPU then yes there wont be much overhead, but then again there will not be much virtualization
Basically a smaller T chip (8 vs. 16 cores) and better threading with out of order and trying to re-announce the sparc version of Exa-crap cause no one would ever spend $20M for a sparc version
Thanks but we will stick with x86 and power for our infrastructure, pity the fools who have not seen the light and still buy this crap.
The only thing interesting is Oracle is finally bringing virtualization to the exadata solutions after almost 3 years...they really need to get a solution that can compete with vmware and powervm
XXOOXX
Power 6 didn't do out of order execution either (http://en.wikipedia.org/wiki/POWER6), so cut back on the poo-pooing a tad. I will admit that the Power 6 ran at 5GHz, though... The point of the T chips was to make them simple which meant they could run cool on low power; the more complicated you make the chips, the more transistors you need and the more power. As for clock speed, ~3.5GHz seems to be the fastest people can make CPUs run at these days where OOO etc is enabled; I remember 3GHz Xeons about 8 years ago in my previous job and Intel are still basically producing 3.0-3.5GHz chips, just with more cores/threads.
The single thread performance does need to be matched against Power 7/x86 to see how it compares, but I haven't seen any SpecINT results relating to that.
Power 6 didn't do out-of-order execution, but Power 5 did, as does Power 7, so Alison has some justification about claiming it not being new.
IBM just had to learn what Intel did with Pentium 4, that high clock speeds and deep pipes are not the answer to overall throughput. That and power consumption issues resulted in Power 7 being a different processor than Power 6.
The reason why Power 6 did not do out-of-order execution was (as far as I am aware) a result of IBM pushing the clock-speed.
I am glad to see that there is still someone other than IBM investing in non-Intel processors. The world will be much more boring if/when x86 becomes the only show in town.
It will be interesting to see how independent comprehensive benchmarks show these systems vs. Power 7 and Power 7+ and the current crop of HP systems, not just the cherry-picked "World Record" results that Oracle put in the announcement. Not that Oracle are doing anything different from all the other hardware manufacturers in their marketing spiel.
Its great to talk (down) the SPARC T4 speeds and feeds but what really matters is the proof-and its definitely in the pudding. The way I look at it, SPARC T4 has just surged to the head of the performance pack compared to Power7 and Westmere. The SPARC T4-4 server is 22% faster than the 8-socket IBM POWER7 server with the same number of cores and has over twice the performance per socket compared to the IBM POWER7 server while achieving 33% better price/performance than the IBM POWER7 server on the TPC-H @ 1TB Benchmark. http://bit.ly/p632rJ . The SPARC T4-4 Server also set a world record on the SPECjEnterprise2010 benchmark, beating IBM POWER7 and westmere x86 on several metrics. http://bit.ly/p637Bx. Theres 7 other world record benchmarks that have been achieved so time to get the facts straight. http://bit.ly/ooRUPA
LOL @ the Sunshiner quoting vendor benchmarks, when anyone with a clue knows those mean nothing in real environments. Dare we ask was the real work in those benchmarks being done on umpteen hundred short-stroked SSDs, in a wildly unrealistic partitioned database? And you can stop laughing there, IBMers, because IBM are just as bad with their benchmarks. At best, they are all just a pointer to possible performance, and should be treated with a healthy does of scepticism. Just try getting any vendor to guarantee that you will see the exact same perfromance, they'll all ryun a mile at the suggestion! First rule of real benchmarks - if it ain't in your environment, with your stack and your data, and on your LAN and SAN infrastructure, it's not worth the paper it's printed on. I always recommend you should try it before you buy it.
Actually, I do agree with you here, Matt. Benchmarks say nothing really.
But, at least I do provide some benchmarks and some support for my claims. Others, for instance Allison Park just spews out lot of claims with no support what so ever. Which means they are all made up. FUD in other words. If Allison could support the claims, I would say nothing.
Same old, Allison, same old. You spew out lot of statements without backing them up. You are basically spreading false evil rumours - that is called a FUDer. If you were able to back your claims up with white papers, benchmarks, etc - then I would say nothing. Because then you were saying FACTS, and not FUD.
Dont you ever get tired? If you at least could back up your claims, then I did not have to counterpost to prove you wrong. I will now post some benchmarks which shows that you are wrong, and as a reaction to my dispelling of your FUD - I will get lots of complaints for defending Oracle. That is my guess. Anyone wants to bet against? (Complaints has happened almost in every thread I had to dispel some IBMers FUD about SPARC and Niagara). Dont IBM supporters ever get tired of all the FUD? That only makes me post benchmarks to counter the FUD. Look: no FUD - no counter benchmarks from me. Is that so hard to understand?
I think I need to end this post with talk about AIX getting killed, about POWER7 is getting killed, etc - just to tease Allison. Allison, you really do want me to go on about AIX and POWER7, right?
.
.
Let us begin countering your wishes, Allison.
"...Shared 4MB L3 cache - 4MB/8 cores has anybody done the math on Westmere or Power7 they should be ashamed of trying to be proud of this..."
As people have told you numerous times, Niagara cpus dont need a big cache to be fast, because of their unique design. How else can you explain several of the world records that T3 has, besting the POWER7 and x86? That would be impossible if Niagara was cache starved. I dont know how many times I asked this question. It is getting really really tedious. Also, the big IBM Mainframe z196 has 300MB of cache, but it is even slower than a decent x86 cpu. Ergo, cache size is not that important. Niagara is fastest in the world in some benches, with a tiny cache. IBM z196 Mainframe 5.26 GHz cpu is really slow even though it has almost half a GB of cpu cache.
I guess this is not the last time I have to try to explain this to Allison Park and other IBMers. Getting really tedious.
.
.
"...."up to 5X the single threaded performance" I am guessing that is compared to T1? T2? can't possibly be compared vs. Xeon or Power cause I am betting it is 1/5th
The sparc Supercluster idea is pure marketing BS Peoplesoft and JD Edwards performance improved up to 8X? I didn't even think anyone had JDE on SPARC. I thought they killed that before Oracle bought Sun..."
Then you still try to imply that Niagara cpus are slow, even though I have numerous times posted world winning benchmarks. Some people never learn
Regarding JD Edwards performance that you talk about, I have posted benchmarks on this too. The T cpu is the fastest in the world, if we talk about JD Edwards benchmarks:
http://blogs.oracle.com/BestPerf/entry/20110701_sparc_t3_1_jd
We see that T3 Niagara cpu is 25% faster than IBM POWER7, and it is 10x faster than Intel Xeon x86 model X5570. I think it is funny that a 1.65GHz with a tiny cache can be faster than 3.55GHz POWER7 with big cache, and win over 2.93GHz Intel Xeons with big cache. Again, cache size doesnt mean much if you know what you are doing. It is only IBM that has fooled you that cache size is everything. If you are good enough, you can be worlds fastest with a tiny cache. Just like Oracle.
.
.
"...Thanks but we will stick with x86 and power for our infrastructure, pity the fools who have not seen the light and still buy this crap...."
You do that. The T3 is apparently 25% faster than POWER7 in the above benchmark. And you need 10 Intel Xeon X5570 at 2.93GHz to match ONE SINGLE T3 at 1.65GHz with tiny cpu. So, who has seen the light? Who is buying the slower gear? Let me tell you: it is you.
Regarding Peoplesoft benches, Oracle also holds world records even there. Just look at the link I provided.
.
.
Regarding the POWER7. I must repeat this many times, because obviously Allison Park is a slow learner, it does not matter how many times I explain something to him, he still dont understand. So, Allison, just for you, here comes an explanation again.
You do happen to know that IBM POWER6 were several times faster than x86? And POWER6 costed 5-10x servers more than x86 servers
You also know that POWER7 is only 10% faster than x86 Westmere-EX in some benches? And POWER7 costs only 3x as much as x86 servers.
So, what do you say about POWER8? It will be slower than x86? Then it must have lower price. And IBM walks away from low margin business, everyone knows that. This means POWER will be killed, when x86 is faster and cheaper. Also, IBM has officially said that AIX will be killed and replaced with Linux:
http://www.zdnet.co.uk/news/application-development/2003/01/29/ibm-linux-will-replace-aix-2129537/
Allison, better start to learn Linux. Your POWER servers and AIX will be killed off by IBM. Because the x86 are getting fast enough, but cheaper. When that happens, maybe then the FUD from you will stop? In that case, I will rejoice. :)
Kebabbert...my point is that Oracle keeps making these big 3x/5x/8x claims but no real facts behind them.
Thanks for posting that 2003 article again and pretending it is relevant to todays market.
Here is the quote from the article that you hope people don't read. I would say you are between a rock and a hard spot but larry killed your rock. :-x
Bowen, the executive in charge of AIX, emphasised that IBM's Unix isn't being replaced by Linux on any product plans. "We've got people now who are building chips for 2007 systems. If we had any belief that AIX was going to fall down and stumble, we wouldn't be doing that," Bowen said. In particular, a major revamp of AIX is due in early 2004. And though Bowen wouldn't provide specifics, he said the AIX development team is somewhere between two and four times as large as the 250 people IBM employs to improve Linux at its Linux Technology Center.
as always nice watching you write an epic fail
"...Thanks for posting that 2003 article again and pretending it is relevant to todays market..."
You are welcome. I will post that article as long as you are posting the same old FUD, that I have dispelled as many times: "Niagara suffers from a small cache - it can not be fast with such a small cache", etc.
Regarding if the article is relevant to todays market, yes of course it is. First of all, IBM said that AIX will be killed in the future, not in the next year. I believe IBM was refering to when x86 is catching up on POWER7. Cheaper AND faster - who will buy POWER7 when you can get more performance for a lower price? IBM has anticipated that trend and IBM knows it will happen, x86 will surpass POWER. Until then, IBM will continue to sell POWER servers. Why should IBM stop making money when they can sell POWER7?
In the future, x86 will catch up on POWER7, the trend is clear. It will happen. Then, is the time AIX will be killed off, together with POWER. Do you dispute the fact that POWER7 is only 10% faster than Intel Westmere-EX? IBM had never believed that x86 would catch up on POWER this fast! Unbelievable fast trend.
.
.
Then you cite some IBM executive yourself. But as you said, that article is old, and IBM did not anticipate that x86 would catch up on x86 this fast. IBM thought they still had some time before x86 catched up on POWER. But time is running out, x86 is in the heels and snapping. Thus, the x86 trend has accelerated much than IBM thought.
Questions to you:
A) Do you dispute that IBM walks away from low margin business? True or false?
B) Do you dispute that POWER7 is cheaper than ever, cheaper than any other POWER generation?
C) Do you dispute that x86 is closing in on POWER?
D) Do you dispute that x86 will be faster than POWER very soon? According to Intel, the Ivy Bridge version will be 40% faster than Intel Westmere-EX.
Can you answer these questions, dear Allison? All your answers will be affirmative. Period. Your answers mean the death of POWER and AIX. And then there will be no more FUD from Allison. Then I will rejoice and drink a toast to you. :)
The zdnet article is dated 2003. 8 years on, AIX is still IBM's main Posix OS, and is unlikely to be completely replaced as long as paying customers want it and IBM makes a profit. And I believe that IBM is making quite healthy profits from their microelectronics division, which produces almost all of the Power(tm) processors that the world uses.
I don't doubt that the proprietary UNIX systems market will continue to shrink, and I don't think than any of the remaining players (IBM, Oracle and especially HP) are really interested in putting large amounts of money into further developing Genetic UNIXes (although they are all pretty well developed as they are).
Linux still has a way to go (IMHO) before it can match HP/UX, Solaris and AIX in overall manageability. I keep expecting to see some major announcement from a large vendor about their Linux distro being as good as the proprietary UNIXes, but I have yet to see it. I am beginning to think that people like Red Hat and SuSE (Attachmate?) are all still in the small system mentality, because of the current in-vogue push for virtualised smaller OS instances running on big systems. Or maybe, the fact that Linux is an Open Source OS means that there just is not the money in it to make that final push into the critical systems market that UNIX currently occupies.
I'm sure I don't know. All I know is that I prefer UNIX (and Linux) to the Microsoft alternative, but maybe I'm just getting old.
BTW. I've often wondered whether there is a degree of jealosy in your comments. Generally, UNIX jobs are still better paid than Linux ones. Are you just wishing for proprietary UNIX to disappear to bring UNIX people down to the level of Linux wages? Just speculating.
As long as IBM is making profit from POWER, IBM will continue to sell POWER. That is true. But today the POWER7 is only 10% faster than Intel Westmere-EX on some benchmarks. And POWER servers are only 3x as expensive as x86 servers. Dont you think that x86 will catch up on POWER soon? When that happens, dont you think IBM has to lower the price on POWER even further?
.
Regarding if I am jealous on AIX and Linux. Well, you dont know too much about me, that is for sure. I do not consider AIX interesting from a technical viewpoint, mostly copying from others nowadays, for instance Solaris. The AIX big days are over. Development pace of AIX has slowed I read from an article here 2-3 years ago.
I do consider Linux inferior to AIX, though. Everybody does, I hope. But of course I prefer Unix to Linux. At my large finance company, we are migrating away from Solaris and VMS to Linux. Which sucks. But it is policy from highest management, and nothing I can do. But I like open source. But Linux is clearly inferior.
and in my view, it's 6 of one and half-a-dozen of the other. Both are still innovating, but neither are doing as much as they used to.
I believe that GPFS is going places faster than ZFS (the actual rate-of-change is staggering at the moment with de-clustered RAID and it's deployment in SONAS devices), but I agree that DTrace was a real innovation. And each vendor has copied part of their virtualisation technologies from the other. IBM is tending to concentrate on technologies that are layered on top of AIX and other OSs, rather than extending just AIX. Whether this means that AIX is stagnating is a moot point, but if the OS is mature and does what is needed, why change?
As to the comment about wages, yes, I do not know you, except by the sometimes over-zealous comments you post here. I was just speculating (in a provocative way, I admit) why you are so vocal about shouting down any UNIX technology other than Oracle/Sun's and Intel's processors. I believe that most people would regard many of your comments as being overly negative.
I've said it before, and I am quite prepared to say it again. There is really no one UNIX vendor at the moment who is 'better' at all things. Each has their strengths and weaknesses. I am glad to see Oracle has not totally abandoned their user base, and hope that they will actually continue to put resource into keeping Solaris relevant, in the same way that IBM is with AIX. I was worried when Oracle were so quiet about Solaris and Sparc after the takeover, and actually began to think that they would quietly relegate the technology to legacy status, but happily that appears to not be the case.
I am not an AIX bigot (at least I don't think I am - comments welcome), it is just the OS that I earn my living supporting, and the one that I know best. When I see something I believe is untrue, I will comment on it, and I will point out areas where I believe AIX has relevant technology where other OSs show deficiencies.
I don't go out of my way to try to put down other UNIXes, and I get annoyed at those that do. The remaining market is fragile enough as it is.
"...Both are still innovating, but neither are doing as much as they used to..."
The fact is Solaris 11 is the biggest step forward, ever for Solaris. It has lots of innovations, ZFS, DTrace, huge scalability, etc. You call DTrace for innovative, that is true. ZFS is also innovative. You know that ECC RAM protects against data corruption in RAM, well, so does ZFS protect against data corruption on disk. I dont know if GPFS has protection against data corruption? And I can assure you that ZFS has many more users than GPFS. In my opinion, it is quite bad to not use ECC RAM, or to not care about data corruption on disk.
What has AIX done lately? Well, it had to be rewritten last year, because it could not handle 256 cores in P795. So now AIX can actually handle 256 cores. Well, Solaris is now targeting servers with 16.384 threads. We are not talking about a few hundreds of threads. Solaris has scaled to hundreds of threads years ago. So what I see AIX is doing, is removing limits it has. I would hardly call it innovation. Sure IBM did lot to virtualisation long time ago, yes. But innovative today? No. IBM only copies Oracle / Sun.
For instance, IBM mocked Sun and said that many lower clocked cores were stupid, and the only way forward was cpus with 1-2 cores exceeding 5GHz - because databases like strong threads. And look at POWER7; many lower clocked cores. Why is not the POWER7 constructed with 1-2 cores at 7-8GHz? No. When Sun created 8-core cpus everyone thought it was crazy, but now everyone has them. So Oracle/Sun shows the path, and IBM and HP follows. And on the way, they mock Oracle/Sun at the same time, while copying. But imitation is the sincerest form of flatter, they say.
.
"...Whether this means that AIX is stagnating is a moot point, but if the OS is mature and does what is needed, why change?..."
If IBM shares your view point, there will be no more innovations from IBM. Why be content? Why be hungry, always looking for improvement? Sun's engineers asked themselves: everyone is trying to make smallest computers, why dont we do the opposite, the largest computer? And Sun created and sold the BlackBox, a container full of servers. Now recently, IBM and HP is also selling... guess what? If you dont see the point of trying to be better, well, what can I say....
.
"....I was just speculating (in a provocative way, I admit) why you are so vocal about shouting down any UNIX technology other than Oracle/Sun's and Intel's processors. I believe that most people would regard many of your comments as being overly negative...."
Of course my comments about AIX are negative. But have you asked yourself WHY my comments are negative? I have explained that numerous times, but probably you missed that. Here is the explanation again: there are lot of IBM supporters here that FUDs a lot, spreading false negative rumours about Solaris and HP. As a counter, I give it back, but the difference is that I do not FUD nor spread false rumours. No, I am only quoting IBM. I dont like FUD and false negative rumours, so I do not do that. Of course I am speculating, but I am very clear with that. I do not write things as "I talked to IBM CEO and he assured me that IBM is going to kill Mainframe business, so you all better migrate before it is too late!" - which would be a false statement and a lie. You see such statements frequently from IBM supporters. To counter all that, I give it back. But with real quotes. I do not make anything up. I have nothing against facts such as negative quotes from Larry, nor do I have anything against SPECULATION, when the poster is clear with that he is speculating. Anyone is free to speculate, but make it clear then.
"Opinions are never wrong. Facts can be wrong". When I see FUD and lies, I correct them. And everyonce in a while, I do to the IBM supporters, what they do to Oracle and HP: give it back.
Why dont you question all the IBM FUD going on here? Why are questioning my "give back"? Without all the IBM FUD, I would not have to counter. I post as a RE-ACTION on what Allison Park and other FUDers say. I react on those posts. Without such posts, no reaction from me. It is as simple as that.
So, will you start to question the IBM FUD going on here, or will you only question my dispelling of the FUD?
If Jesper and Matt are spreading FUD, then they are doing it in a way that is less rabid than you. I would be surprised if you are not foaming at the fingertips when you type some of the things you do.
A point in question. How much re-writing do you think is necessary to increase the number of processors managed by an OS. According to you "it had to be *rewritten* last year, because it could not handle 256".
Well. All that is really required is to change a couple of numbers in the kernel header files, and re-compile the kernel and any tools that reference those headers. In fact, the support was included in a PTF fix for AIX 6.1, not even a new level. Definitely not a re-write, more like a tweak.
Like other shortcomings, I guess that you have never worked in source at a kernel level for a UNIX, and I would also hazard that you never had to play with sysgen-ing an older SunOS release. Honestly, the more you say, the less relevent what you say becomes.
When it comes to new OS features, what do you think that Oracle are adding to Solaris 11? Both DTrace and ZFS are old news. How often can you consider them new (both have been around in previous versions of Solaris), and neither of them are really an extension to the OS. They are what IBM would call 'layered products'.
Unlike Linux and Windows, the remaining UNIXes, and especially AIX IMHO have a mature set of APIs, RAS features and other management processes. I will concede that lack of change may indicate stagnation, but excess change may also indicate immaturity and feature bloat driven by marketing hype. There is middle ground. What new features would you like to see in a UNIX?
On the filesystem front, ZFS moves the disk hashing up into the filesystem layer, and produces protection at the file or other disk object level. Reed-Solomon encoding of data at the filesystem block level effectively does the same in the GPFS de-clustered raid system. Big deal. And apart from Sun themselves, not everybody believes ZFS is safe. See this paper www.usenix.org/events/fast10/tech/full_papers/zhang.pdf that was presented at Usenix, which concludes that ZFS may be more tolerant of disk errors, but is not invulnerable to data corruption.
There is a fundamental design difference between the T series Sparc processors and the Power series of processors. One that is closing from both ends, and again they are converging on the middle ground. The announcements of what was it - heavy thread?- just shows that the Sparc design is being changed. One of the real problems with T1 and T2 processors is that they were committed to the lightweight thread model, which made them excellent for many small processes, but very poor for smaller numbers of large processes. Why is this change an innovation, and IBM putting a larger number of slower cores a realisation of a deficiency. I believe that Matt described this far better than I can, elsewhere in this thread.
I think I agree with Jesper's analysis of Larry's announcement claims. They look good, but do not actually stand up to any real scrutiny, as they claim things that other vendors do not bother to benchmark, or cannot with the same levels of code. It's like you saying that you are the fastest person on earth at running from your front door to your living room, but you never allowing anybody else in to your house to try and beat your time. Surely you can see this?
So, please calm down, and actually read stuff that comes from people other than Oracle's marketing team.
"...If Jesper and Matt are spreading FUD, then they are doing it in a way that is less rabid than you..."
As long as what they say have bearing in reality, it is not FUD. It is not a lie. I suggest you read the definition of FUD on wikipedia.
Regarding if Jesper and Matt is spreading FUD, and they do it in a less rabid way - so you admit they are spreading FUD? :o) But actually, Jesper is not a real FUDer, as he backs up his claims, and he i knowledgable. Matt, on the other hand, has numerous times spread FUD, such as "T3 suffers from a small cache, and can not be fast" - well, the T3 holds some world records, how can that be possible if T3 is not fast? Something is wrong, either the T3 does not hold world records, or Matt is wrong.
I hope you dont imply that I spread FUD when you write "they spread it in a less rabid way" - less rabid way [than me spreading FUD] or what? Have you seen me giving unsubstantiated claims, without being able to back them up? No. Hence, I am not spreading FUD.
How about you? You seem to defend the FUDers here, and do not object against all the FUD going on. There are lot of FUD going on here, yes. But I do not see Oracle people spreading FUD, no Oracle people are like Allison Park or Matt Bryant. They are mostly quiet. But I dont work at Oracle, so I can say what ever I want. :o)
When I come in and dispel the FUD by posting benchmarks and white papers - you do complain. Not on the FUDers, but on me. And you seem to be clearly AIX biased, as you get upset on the IBM official statements about AIX being killed. Why are you complaining on people trying to dispel the FUD? Some people could draw the conclusion that you are an IBM supporter, pretending to not be.
I dont get it, Peter. If someone gets bullied and defends himself, why do you complain on the defender? Why not the attacker? Clearly biased? I would not be surprised if you worked at IBM.
.
Regarding your talk about GPFS, no, you clearly dont understand. And, at the same time, you are attacking my credibility with:
"...Like other shortcomings, I guess that you have never worked in source at a kernel level for a UNIX, and I would also hazard that you never had to play with sysgen-ing an older SunOS release. Honestly, the more you say, the less relevent what you say becomes...."
I think it is funny. You dont know what you are talking about, and attack my credibility, implying that I dont know as much as you do. I do not talk about all my different university degrees, and what I have done, etc - that is very cheap to do. Of course I could break you with that, but I dont. It is too cheap. Instead, an debate is not wun with "I know best, I have this degree" as you think. I suggest you change your debate strategy. Use good arguments instead of attacking the other person. Maybe he is more well educated than you, why focus on the CV?
You started off nice, but it did not take many posts before you revealed yourself as an IBMer in disguise. GPFS some R-S codes doesnt help to protect data. Did you not know? Do you believe that ECC RAM protects against data corruption?
.
Then you start spreading the FUD:
"...And apart from Sun themselves, not everybody believes ZFS is safe. See this paper www.usenix.org/events/fast10/tech/full_papers/zhang.pdf that was presented at Usenix, which concludes that ZFS may be more tolerant of disk errors, but is not invulnerable to data corruption..."
Well, have you read that research paper? What is the conclusion? That ZFS DOES protect against data corruption. You are claiming the paper gives the exact opposite conclusion - hoping that no one would care to read it. But that paper gives support for ZFS ability to protect against data corruption. I suggest you read that paper. Thank you for further supporting my claim that ZFS gives good protection against data corruption.
No, it says that it protects from DISK ERRORS, and I said that.
My education is not great, just an ordinary degree. My 30+ years of UNIX, much of which is on kit other than IBM (including Sun, Digital, HP, Data General, Amdahl, AT&T and many others over the years), and a good part of which has involved UNIX source is more relevant, although if you look, I have often said that I am currently contracting for IBM, and that in the past I did work from them. I I am no marketing shill, however. I just appreciate features that make the work I do easier. I am as critical about some of the features as anyone else, and I often get very worked up by their support processes. When IBM entered the Open Systems world in 1990, they were regarded as the Big Enemy by many UNIX people, myself included, but I think that they did actually prove themselves.
Recently, whenever I have had to do work on HP/UX and Solaris, I have found that it is less easy than AIX, but that could be because I am more familiar with AIX. I definitely feel that Solaris and HP/UX feel more like traditional UNIX systems than AIX when it comes to management. Maybe a good thing, maybe not.
I don't know what you think I don't know? I am perfectly prepared to admit that I do not know everything, and I can be wrong. But what have I said that was wrong. The ZFS paper is quite clear, as are it's conclusions, which I referred to. I know GPFS enough to know that what I said was correct. I have compiled kernels enough to know what is involved there.
In relation to what I said regarding Fear, Uncertainty and Doubt (yes, I know that without looking it up in Wikipedia, I have been using the term myself for about 20 years), the first word in my sentence was *IF*. I did not acknowledge that what Matt and Jesper say was FUD, although I definitely would categorize some of what you say as such. In fact, I think I agree with Jesper on almost everything he says on these comments. Very detailed analysis, and worth reading.
I apologise for resorting to ad-hominem arguments. It's always a poor tactic, but sometimes what you say is not thought through, or maybe seen through a filter. I definitely know that what I say is often coloured by my experience, so maybe I should accept that it will be for everyone, but it may be worth you re-examining what you say sometimes. It comes across as very shrill.
You are like one of those kids in the playground who shout and insult everybody around, and who then claims that they are being bulled when one of the people at the receiving end is finally annoyed enough to put you in your place.
Damn, resorting to personal insults again, but you make it so easy!
"anyone else, and I often get very worked up by their support processes. "
Yes, the worst response is still "Works as designed". And the whole 'We are measured on closing, not necessary solve, as many problems as possible, that seems to be the common trend in the industry, also makes blood preasure heat up.
"When IBM entered the Open Systems world in 1990, they were regarded as the Big Enemy by many UNIX people, myself included, but I think that they did actually prove themselves."
Isn't that a bit of an understatement ? I do remember how I felt myself, after growing up on VAX running BSD, HP9000 systems also running BSD and motorola 680XX based SUN systems running Sunos. I remember my first real UNIX job in the mid nineties, where I desperately tried to get my employer to use HP systems rather than RS/6000 machines, until I actually kind of started acting like a professional and gave it a chance. And AIX was actually a fresh breath of air. LVM was clearly superior than the rather arcane ways of many other UNIX'es, and SMIT was a nice easy way to get started on the OS. But it took some years to get rid of the heretic label, that my friends who most of them worked with Solaris put on me.
Today where I have very little hands on, that's the price you pay for going the IT-architect way, my perspective have changed, today it's all about money, and less about technical 'stuff'. But I have found that enabling technical excellence is perhaps the best way to save money.
// jesper (The BSD devil for old times sake)
Well, nice of you telling us about your background. But I dont really thinks it is relevant here. If I told you of how many university degrees I have, at top universities and what I work with in a large world known finance company - so what? Does that make my arguments better?
No. I think such reasoning is pure snobbery: "My family is wealthy and are noble, you should call me Sir - and I have no clue of this subject, but what I say counts. Because you are poor, you dont count". So what does that have to do when you debate? It has happened that professors (yes, real professors) have said something really weird, which I have totally rejected. And then there are uneducated persons that have said clever true things which I have accepted. What people has done or not, is irrelevant. So I think it is of not relevant when you attack my Unix background and that I have no experience of AIX, etc. I will not discuss that further, as I will not question your education. Your view point still counts to me, no matter your education (as long as you dont lie and FUD).
.
Regarding your statement about ZFS:
"...not everybody believes ZFS is safe. See this paper www.usenix.org/events/fast10/tech/full_papers/zhang.pdf that was presented at Usenix, which concludes that ZFS may be more tolerant of disk errors, but is not invulnerable to data corruption..."
No one claims that ZFS is 100% safe. Of course there are bugs in ZFS and people have had problems - just like every complex piece of software.
But I claim that ZFS gives better protection than most, because ZFS is designed from ground up to protect against data corruption when it occurs ON DISK. ZFS is a file system and protects data on disk.
Then you clarify:
"...No, it says that it protects from DISK ERRORS, and I said that...."
What do you mean with that? ZFS protects against disk errors, of course. What do you mean when you say "not everybody believes ZFS is safe, it protects against disk errors but does not protect against data corruption"? Do you mean that ZFS protects against disk errors, but not against faulty RAM sticks? Do you mean that ZFS to be safe, should be able to repair faulty RAM memory sticks? Could you clarify further what you mean?
.
Regarding GPFS, it does not matter how many kernels you have compiled. If you dont understand the theory behind R-S codes, then you it doesnt help you. R-S codes are not safe. If you think that, you are wrong. I am convinced that ZFS protects better than GPFS - but I have not read any academic studies on a comparison.
.
"...I apologise for resorting to ad-hominem arguments. It's always a poor tactic, but sometimes what you say is not thought through, or maybe seen through a filter...."
Apology accepted.
.
"...I did not acknowledge that what Matt and Jesper say was FUD, although I definitely would categorize some of what you say as such. In fact, I think I agree with Jesper on almost everything he says on these comments. Very detailed analysis, and worth reading...."
If you categorize some of my sayings as FUD, I would really like to know what. If you can quote me FUDing, I will of course stop say that. I do not want to lie, mathematicians really detest lies. So go ahead and please quote me where I FUD and lie. If you can do that, I will not say those things again. I think it lies in your interest that I stop FUDing, so please quote me.
.
Regarding Jesper and Matt. Jesper dont FUD and he is no liar. But he is clearly very biased, although knowledgeable. I remember when I showed a benchmark where Sun won (had higher throughput), and Jesper proclaimed "it does not count as IBM had lower latency". Later I showed a benchmarks where Sun had lower latency, but IBM had higher throughput and Jesper proclaimed "it does not count as IBM had higher throughput". So it does not matter what benchmark I show, something will always be wrong. This is clearly bias.
Regarding Matt, he FUDs sometimes. He talks about the low performance of Niagara cpus and how cache starved it is - and everytime I show benchmarks where Niagara is fastest in the world. How can Niagara be cache starved and slow if it holds world records? Still Matt repeats himself. Although he has been target of IBM FUD as well. I remember the Itanium article, it was full of IBM FUD. I dont get it, why are IBMers attacking and FUDing everyone? HP and Oracle and I dont know who more?
.
"...When IBM entered the Open Systems world in 1990, they were regarded as the Big Enemy by many UNIX people, myself included, but I think that they did actually prove themselves...."
This is strange, how did IBM prove themselves? I have many counter examples. Such as when IBM released 511 patents to open source community, and TurboHercules (writes IBM Mainframe software emulators) got attacked by IBM for using those patents. The problem is that TurboHercules is really fast on x86 and rivals Mainframes. Or, actually the problem is that Mainframe cpus are really slow.
You want me to go on and show some examples of IBM FUD and articles? And then you can explain how IBM proved themselves? According to wikipedia, IBM was the first company that employed FUD systematically to discredit competitors. IBM has never stopped FUDing, and neither have their employees, just look at the threads here. Full of IBM FUD targeting everyone.
If AIX ran on Intel, then you would probably have a point, but the market for POWER is "AIX on POWER" and "IBM i on POWER". It's the whole package.
As far as I am aware, IBM is making no effort to (re-)port AIX to an Intel architecture and as far as I am aware will never consider i on that platform (I say re-port, because AIX 5.1L was available on Itanium, although nobody was interested in buying it). Earlier versions of AIX were also available for i386 and i486, but only on IBM PS/2 systems (in the same way that SunOS 4 was available on the Sun 386i system in the late '80s), but that was actually a different code tree (AIX 6 and 7 go back to AIX on the 6150 PC/RT platform, whereas AIX PS/2 came from the IX and AIX/370 mainframe port, originally done by Locus). There were big differences, and the only people who made any attempt to treat them as the same OS were the IBM marketing people.
I'm sure that a port could be done (it's almost all C anyway) but as you've pointed out, if POWER were to be dropped as a platform, I'm sure that IBM would produce an enhanced version of Linux with some form of AIX compatibility/migration tools to try to keep their customer base rather than port AIX. But that is not on the cards at the moment.
Strangely, out-and-out performance is not the primary motivation for large AIX customers to keep buying it. AIX itself and the RAS of POWER systems are, along with the associated support skills that the customers have invested in. If you are involved in a move from VMS and Solaris to Linux, I'm sure you are aware of this. The move from VMS is obvious (where else are you going to go), but I wonder whether the move from Solaris is because of lack-lustre statements of intent from Sun/Oracle with regard to Sparc and Solaris. If that is the case, then this latest release from Oracle is too little too late, at least for you, and I sympathize.
"...AIX itself and the RAS of POWER systems are, along with the associated support skills that the customers have invested in...."
I agree that AIX has far better RAS than Linux. This is true, and I can not deny this. It is a fact. However (just to tease Allison Park and his FUD), do you agree that
1) x86 is catching up fast in terms of performance?
2) Intel is adding RAS to Intel Xeon cpu quite quick
3) The price difference between x86 servers and POWER7 servers is smaller than ever?
4) IBM has officially said that they are going to replace AIX with Linux, in the future?
Is everything I say true, or false evil rumours that have no bearing in reality? Which one of the above statements are false?
Compare all my true statements, with the FUD and lies from Allison Park, with no links nor supporting evidence.
How come you are not questioning Allison Park with his obvious FUD? Why are questioning me when I dispel Allisons FUD? Why are you questioning me when I post true statements? I dont post false statements, do I?
"1) x86 is catching up fast in terms of performance?"
Nahh... I caught up quite a few years ago. The problem is that the field seems to be narrowed down to INTEL x86 and POWER, when you talk about 'performance'. The rest including SPARC is pretty much looking like legacy and niche products.
"2) Intel is adding RAS to Intel Xeon cpu quite quick"
Jup. And the price is rising. But Xeon -EX'es are really the sweet spot for Intel. The highend x86 marked is tiny compared to for example the rest of the highend marked.
"3) The price difference between x86 servers and POWER7 servers is smaller than ever?
Jup. POWER7 rocks.
"4) IBM has officially said that they are going to replace AIX with Linux, in the future?
Yawn. IBM also said once that only 4 computers were needed in the world. Come on.
// Jesper
LOL, I was just enjoying Alli's biting response! I'm surprised she missed the opportunity to point out the new T4 systems only scale to four sockets compared to Pee7's Power 795 offering. But then that would be rather mean of her, comparing a webserver to a real enterprise system.
Seeing as TPM as has done his usaul and compared only to IBM (yeah, I'm trying to look surprised), I'll do the quick compare with hp's Integrity.
T4-1 vs rx2800i - In the same 2U, hp puts in two sockets to Snoreacle's one. And those are real CPU sockets, quad-core Tukzillas, not Niagara wheiner CPUs rehashed, so they have real single-threaded performance. They also are ready for the next gen Poulsen CPUs which will be octo-core and will be compatible with current hp-ux apps unchanged. The Snoreacle idea of compatibility with old Slowaris apps is a long list of "but you must have this version".... And what other OSs does the T4-1 support other than Slowaris? Memory? Well, the rx2800i has 24 memory slots and can scale now to 384GB of RAM, 50% more than the T4-1. So far, the Snoreacle option doesn't look like much of an option at all.
T4-1 vs BL860c i2 blade - again, hp has two sockets on the base blade, ready for the next gen octo-core Poulsens right out of the box. Who knows what is coming next with the SPARC roadmaps, but expect new boxes for any future generation just like there have been for each existing generation. Again, the hp solution has 50% more RAM capability. In the space you need for five of the 2U Snoreacle boxes you can fit a C7000 chassis and eight BL860c i2 blades, plus associated KVM, SAN and LAN switches, making the Snoreacle option more than a trifle space-inefficient. I'd try making a comparison between the Snoreacle blades and hp's but what's the point, who buys Snoreacle blades?
T4-2 vs BL870c i2 blade - Snoreacle gets up to two sockets and 512GB of memory! But the hp double-blade now has four sockets and 768GB (yes, 50% more again - hey, Larry, don't databases like Oracle like lots of memory?). If you can find a 1U CNA switch, and just ignore the KVM and redundant switch options of the C7000, then you could try three of the 3U T4-2s in the same space as a C7000 with four BL870c i2s. Well, if you were a sado-masochist with money to burn, that is. Best not mention the advantages of the Virtual Connect technology that the hp blades can offer.
T4-4 vs BL890c i2 blade - Snoreacle botches two motherboards into one chassis (how? - are they using a switched backplane or just some software to make two separate hardware instances appear as one 8-way server?) and finally hits the dizzying heights of 4 sockets! In 6U. Not compact. That's six in a 42U rack as you're going to need space for switches at least. But six BL890c i2s can fit in 30U, without the need for additional switches as they go in the back of the C7000 chassis. Leaving me with 12U extra space over the possible 2U spare after adding two 1U SAN and two 1U LAN switches to the six T4-4s.
And what do we compare with T4-wise with the Superdome2? Don't expect anything soon, even if Larry manages to stick to his fairytale roadmap! Maybe it would be fairer on the T4s if we compared them to the Xeon-based ProLiants, but then we'd run into the same problems with the lack of scale in the T4 offering - what do you compare to the Giant ProLaint DL980?
Instead of swallowing more Sunshine, I suggest you check out reality:
http://www.channelregister.co.uk/2011/09/23/oracle_second_quarter_idc/
To help you with the comprehension - EPIC (that's Itanium) is up, despite Larry's finest FUD; Snoreacle is dying; and Fudgeitso is looking terminal!
At least I think Matt believes what he says. Anyway...
Matt, you quote an article that only brings up European revenue, which is obviously bad. Both IDC and Gartner show that worldwide Oracle is increasing highend market while HP is decreasing.
You may want to just stay on the floor Matt.
Matt...thanks for the compare.
With last years 2x price per core for 13% more performance of Tukwila (compare HP's TPC-H benchmarks) and this years list of product releases that will not be on Itanium it looks like most HP customers have also given up on Itanium. I have yet met one that is looking to go to the Snorcle kit to reward Leisure suit Larry.
Maybe Meg can sweet talk Larry
".....it looks like most HP customers have also given up on Itanium...." Oh dear, Alli has fallen into the Sunshiner trap of thinking what she wants to believe is what is actually happening. Sorry, Alli, gonna have to refer you to exactly the same article as the Sunshiners:
http://www.channelregister.co.uk/2011/09/23/oracle_second_quarter_idc/
Now, try and concentrate, read the bit ".....EPIC was up seven per cent...." - that's Itanium, meaning hp Integrity. Enjoy!
As I understand it, after hp's lawyers get through showing the agreements that Oracle signed up to in court, Larry will be buying a lot of hp lunches.
Oh, Kebbie, if you had a clue you'd be interesting! Benchmarks produced in lab conditions with wildly unrealistic setups are pointless in the extreme, as I have pointed out to you umpteen times. Consider ThrustSSC, the current landspeed record holder. Nothing is faster than ThrustSSC on the ground. There is no way you would say Thrust could be beaten by a Transit van or an articulated lorry, so ThrustSSC must be "better". You'd probably expect the Transit to be faster than the artic over Bonneville saltflats as well. But, your comparison is only relevant to the "how fast can it go in a straight line" question.
Now, think about having to move a dozen grand pianos from London to Glasgow. Suddenly, the Transit van is a better option than ThrustSSC, which can't cope with public roads let alone carry a piano, and the artic the best option of all. Now, which one do you think is likley to be a more realistic daily requirement, running at Mach 1 in a straight line, or being able to deliver a dozen pianos? The benchmarks you love to quote are all in the ThrustSSC mold - "how fast in a straight line" - and are interesting if artificial technical experiments, but have zero relevance to the real and daily requirements of businesses.
Real daily requirements are going to be things like "how many of my hot database tables can I squeeze into RAM?" - the hp option is better for every comparison. Or the resilience question, "What if I have a problem with a CPU socket?" - having more sockets means more resilience, and again, hp offers better options throughout the range.
In short, forget the benchmarks and other marketeering guff and try doing some original thinking.
There is lot of FUD and BS here, I totally agree with you. However, the FUD and BS is only from the IBM direction.
I am only posting true statements, with supporting evidence such as benchmarks, white papers, research papers, etc if someone requests that. For instance, do you want me to prove that POWER7 is only ~10% faster than Intel Westmere-EX on some benchmarks?
"......Anyone else think that Allison Park and Matt Bryant should be el Reg couple of the year?" Whilst Ms Park does have some dubious taste in server vendors, I think that would presuming far too much depseration on her part. Oh, and Mrs Bryant would probably have a few choice words on the matter too!
First of all, it's nice to see that Oracle have managed to get a decent processor out. That is truely good news for the competitive landscape and the UNIX marked.
Now I would have liked Oracle to actually have come up with some more benchmark results that were in the category, "Now we can really compare to something". Rather than the usual suspects where there either isn't anything to compare against or it's their own 100% controlled software stack with a new version that nobody else is running.
But in benchmarks and war you should only pick the fights you can win. And Oracle is very good at avoiding Westmere-EX and POWER7 where they know they can't win.
Lets have a look.
TPC-H.
This is truly a great benchmark result. It's a real banger. But to be honest I think that Oracle is the only vendor who has truly cracked this benchmark. If you look at all the submissions except the Oracle DB on SPARC then they have much longer run time for Q1, Q9,Q13, Q18 and Q21.
The recent Oracle benchmark on the M8000 and the T4-4 actually runs faster than the HP DL980 with 80 Westmere-EX cores.
Now on a benchmark like SPECINTRATE the M8000 does 882@64 cores where as the DL980 with 80 cores does 2080. So IMHO it's the software stack and perhaps setup that allows for Oracle to excel in this benchmark, not so much the hardware. And what really points toward the setup and that they have cracked the benchmark is that the SUPERDOME 2 submission with the Oracle DB shows the same pattern as the other non Oracle on Oracle hardware. I would have presumed that the SD2 was at least 25% faster than the M8000.
Which kind of makes me better understand why HP is so pissed at Oracle. If HP suspects Oracle not helping out the max, when not doing submissions on their own hardware... well
As for the POWER7 based POWER 780 submission. Come on it's Sybase on Linux, that's not really a serious measure of what the platform is capable off.
jEnterprise2010.
Well, as I have said before this is a benchmark that nobody really heard about before Oracle started using it in their marketing material.
Now there is no denying that Oracle also here have made a good benchmark .
If you compare the setup of the two submission that Kebabbert are pointing to you can clearly see the difference:
http://www.spec.org/jEnterprise2010/results/res2010q3/jEnterprise2010-20100628-00012.png
versus
http://www.spec.org/jEnterprise2010/results/res2011q3/jEnterprise2010-20110907-00027.jpg
But you have to acknowledge that the Oracle submission uses Flash disks, 8 times the number of storage nodes, 12 times the memory, 16 times the network bandwidth, 8 times the storage bandwidth. And last sure they only use x4 the number of cores. But CPU is not always the bottleneck.
Then there is the JDedwards benchmark (http://blogs.oracle.com/BestPerf/entry/20110701_sparc_t3_1_jd)
Get real Keb. You are comparing Oracle running ontop of System i formerly known as OS400, to a UNIX platform. If you knew anything about AS400/os400, you would know better than to deduce anything about Oracle performance on AIX based upon those results. But I guess you don't.
So to conclude nice benchmarks but it's still to cherry picked to really conclude much about the true performance of T4 compared to the other players, but that it's a huge step in the right direction from Oracle, if it turns out that it has a good single threaded throughput. I just don't understand why they had to go through T1-T3 to get to T4. Now compared to T3 I would say that the throughput is most likely the same.
If we look at the jEnterprise2010, result then we will see that the T4 result gives pretty much x4 throughput with x4 the number of nodes.
So Oracle have managed to keep throughput while, speeding up (which is still to be seen by how much) the single threaded throughput. That is great, but IMHO still not good enough.
// Jesper taking on his flame resistant hat.
Fine. I remember when I posted a benchmark when Sun had higher throughput than IBM, and you complained and said something like "It means nothing, because IBM has lower latency". Later I posted another benchmark when Sun had lower latency, whereupon you said something like "it means nothing, because IBM has much higher throughput". So no matter what, Sun could never win.
Never mind about the T3 results. What is your comment regarding the T4 results? Here are some world records, with more down below this one.
http://blogs.oracle.com/BestPerf/entry/20110927_sparc_t4_2_jdedwards
"Never mind about the T3 results. What is your comment regarding the T4 results? Here are some world records, with more down below this one.
http://blogs.oracle.com/BestPerf/entry/20110927_sparc_t4_2_jdedwards"
Are you really such a sucker for marketing material ?
The T4 benchmark that Oracle have run is not even in the same category as the IBM iSeries result mentioned. You can't really compare them.
Furthermore the iSeries result is not an official result of any kind, if you had bothered following the link to the IBM website, it's an internal IBM test of the JD edwards stack, it's not even an official submitted benchmark result.
Lets just say that the Oracle marketing guys must have looked a long time for this one. Geeezes..
And again this is 60GB of RAM and 48 normal disks, with a CPU utilization of 60% running with DB2 as the database (Not a the optimized Oracle stack for JD Edwards) versus a fully native software stack running on 3 machines with x10 the RAM, internal flash disks and 2 flash Storage servers.
Gee, comparing this is.. to put it mildly not serious. If you can't see that .. well.. you have a problem IMHO.
// Jesper
I like things that I can post and prove my claims, therefore benchmarks are quite good. They give a simple number that you can relate to, and they are repeatable which is good.
I dont like claims such as these, by some Anonymous which were very frequently posted earlier:
"I work at a large Bank/Exchange/Telco/whatever. We are using SPARC and I really love them, but now we tried IBM POWER6 and they are so much faster so we are now migrating to POWER6. You do that too, or you will start to loose money
/ex-SPARC lover"
This claim is clearly false, because it was always posted with the same wording, just 2-3 sentences. And always the intro "I work at a large bank" one day, next day "I work at an exchange" etc. Very little imagination.
When I post, I want to post checkable facts, so everyone knows that I do not post false things like IBMers do. That is the reason benchmarks and white papers are good.
Keb. I don't care about IBM/SUN/ORACLE whatever anonymous fanbois that post strange things. Neither should you, and neither does anybody else.
The problem with you is that you echo without any reality check what soever, Oracle marketing material. And the material is pretty desperate at the moment cause to be quite frank Oracle is under enormous pressure in the UNIX marked.
// Jesper
"....Keb. I don't care about IBM/SUN/ORACLE whatever anonymous fanbois that post strange things. Neither should you, and neither does anybody else...."
Well Jesper, you gained a lot of cred in my eyes here. It shows you are critizing FUDers. I wish more people here would be like you, Jesper. At least you know what you are talking about, and post and dissect benchmarks. In my eyes, you are one of the IBMers I have most respect for.
.
"...The problem with you is that you echo without any reality check what soever, Oracle marketing material. And the material is pretty desperate at the moment cause to be quite frank Oracle is under enormous pressure in the UNIX marked...."
I am echoing marketing material because I try to counter FUD similar to "the SPARC T3 is slow because it suffers from a too small cache". To disprove that, I post some world records. That should silence the FUDer. I dont think a viable strategy would be "no, you are wrong the T3 are really fast!" - without posting any evidence. I better just post some hard facts, and then we can see that the T3 is in fact fast - contrary to the FUD.
Regarding these SPARC T4 and T3 benchmarks, it seems you dispute the fact that the T3 and T4 are really fast in some benchmarks. Some believe that T4 is faster than POWER7 in several benchmarks. This you dispute too, as can be seen since you reject the T4 benchmarks because of this or that reason, so you mean the benches have no bearing in reality. Fine. Everyone is entitled to have an opinion.
(I have no problems admitting that POWER7 is a good cpu, and it is fastest in the world in some benchmarks today. And POWER7 would be my top recommendation depending on the client's workload. A fact is a fact. A truth is a truth.)
You go all zombie when you read Oracle marketing material don't you ?
SPECjEnterprise2010
Yes, this is the fastest submission, sure it it is. But if you really wanna conclude anything, then it should be that WebLogic is a damn fast application server. That will run fast when you throw a lot of flash disk and and and... after the benchmark.
Again this is a benchmark that really haven't many submissions.
TPC-H
The oracle submission is... NOT.. I REPEAT NOT the world record, there are several other servers that are not only faster but also a good deal cheaper on the price per transaction.
The Oracle E-Business Suite R12.1.2 XL
Yes, world record.. why.. cause T4 is the only submitted result.
Oracle PeopleSoft Enterprise Payroll (N.A.) 9.1
Lots of results here, but T4 is the only one using the newest version. What is mentioned is that there actually is a mainframe result where the T4 with 32 cores 256GB of RAM and Flash drives and native Oracle software stack, only manages to outperform a older generation IBM z10 mainframe with 9 cores and 32 GB of RAM using DB2 (not a native platform for the APP) which isn't using flash drives with a factor of 2.8. Meaning that the per core performance of the mainframe is 25% better than the T4.
How sad for you Keb, a old mainframe being faster....
Oracle PeopleSoft HRMS Self-Service 9.1.
Whenever I find a link.. it's broken.. so it's kind of hard to actually verify anything. But again it's oracle's own software stack.. in a new version .. bla bla..
Oracle JD Edwards EnterpriseOne
I've covered that one.
The rest isn't really world records or anything just claims about how good the machine is.. Kind of not very serious.
So what was your point Keb ?
// jesper
"...So what was your point Keb ?..."
Nothing really. I quote myself:
"I remember when I posted a benchmark when Sun had higher throughput than IBM, and you complained and said something like "It means nothing, because IBM has lower latency". Later I posted another benchmark when Sun had lower latency, whereupon you said something like "it means nothing, because IBM has much higher throughput". So no matter what, Sun could never win."
I knew that Oracle could never win anyone of these benchmarks. I just wanted to see which excuse you would be using.
Ah, you must be "Allison Park" in disguise? :) I recently posted a benchmark between a cluster of six x86 servers and IBM P795 - which the x86 servers won of course. And guess what? Allison got upset and claimed I "insulted his intelligence" for posting such a benchmark, in a discussion about x86 vs RISC cpus. I never understood why posting a benchmark insulted Allison, but hey, I dont know what is going on in his mind. So, when I post an IBM article and you get upset - are you Allison Park in disguise? Quite probably...
I have no reason to post anonymously.
As far as your questions. I don't need to bother. But Larry is the first person to say he will get out of the x86 server business so what does that mean for exa-crap? Another exadata-1 killing as they move to oem dell or intel directly?
We have x86 and power. The main differentiation is reliability and TCO not a some single cluster vs. SMP benchmark.
Power still commands a pretty high margin and if you count their software stack with db2, websphere, tivoli they probably make $4 for every $1 of hardware. I am sure Oracle enjoys the success of Power also in the market with their insane pricing.
Would be interesting to hear your answers to your own questions in regards to SPARC vs. Intel x86.
And I am not sure what country you live in but I am a she not a he.
"...I have no reason to post anonymously...."
Because you are no stranger to FUD and foul play, it would not surprise me if you posted anonymously too, trying to give the impression that you have a massive support from everyone here, when in fact some people think you should stop FUDing.
.
"...But Larry is the first person to say he will get out of the x86 server business so what does that mean for exa-crap?..."
Larry said he is not interested in the low margin x86 commodity servers. What Larry is interested in, is the high margin, tailor made systems, such as ExaData and ExaLogic - that are built on high end x86 servers. For this he can charge a premium.
.
"...Would be interesting to hear your answers to your own questions in regards to SPARC vs. Intel x86...."
Of course x86 have catched up and surpassed SPARC64 in terms of performance. But SPARC64 has better scalability and for large systems SPARC64 still gives better results in some cases.
However, the T3 and T4 are tailored to another workload. POWER7 and x86 targets the same workload: few strong threads. Not high throughput. Especially T3 is tailored to high throughput, maybe also the T4 (I do not know yet). In terms of throughput the T3 has a definitive and sharp edge. For instance, benches show you need 10 Intel x86 at 3GHz to match one low clocked T3 in some throughput benches. 10x faster definitely counts as a definitive and sharp edge.
Thus, Niagara is a niche cpu targeting throughput and a cpu targeting few strong threads will never reach the same throughput as a throughput chip. In other words, Niagara is not threatened in it's niche, by any strong thread cpu. It would be a miracle if a cpu would have few strong threads and at the same time, have 10x througput just like the Niagara. Thus, Niagara will thrive in it's niche. Maybe T4 is now a general purpose cpu - I dont know. But Niagara will thrive in high throughput niche.
Regarding future of Solaris vs AIX. Solaris runs on both x86 and Niagara. Thus, no matter which workload you have, Solaris will run it. Is it strong few threads? Run x86. Is it high throughput? Run Niagara. Is it both? Run T4(?). Niagara is not competing with x86, but POWER is.
AIX is competing with x86, they are both targeting the same workload. And I do believe that x86 will soon be the king in terms of performance. Now POWER7 is faster. But for the first time in history, x86 will be faster. In a few years. When x86 surpasses POWER, what will happen? Your guess is as good as mine. But maybe, IBM have to lower the price. The race between POWER and x86 will only have one outcome. Look at past historical data and try to extrapolate into the future.
The race between Niagara and x86: Niagara is always faster in it's niche. Both have a place in the market. Few strong threaded cpus: x86 or POWER. Who will win? Massive throughput: Niagara.
I dont really know about T4. Is it more similar to x86 and POWER? In that case, x86 will win in the long run.
Are you refering to my article from 2003? Well, why is that trolling, to post official links to IBM? In the link, I can explain that, IBM says in the future AIX will be killed and replaced by Linux. So when I quote IBM on this, how can I be a troll? I dont get it. Even when I have a source back to IBM itself, I am called a troll? I mean, how much better supporting evidence can I get? It is impossible to have better supporting evidence.
It is like in court, if he confesses the murder and shows the weapon, then how much clearer can it be?
On the other hand, when IBM supporters such as Allison Park spew out FUD and false negative rumours with no supporting evidence - then everything is fair and well, right? If Allison and the other FUDers would have links back to Oracle or Larry at least, but no. They have made up all the evil false rumours. And you dont complain on the IBM FUD.
What is it called? Hippo... something. I think it ends with crasy?
Jesper posts, "I just don't understand why they had to go through T1-T3 to get to T4. Now compared to T3 I would say that the throughput is most likely the same"
To understand the T series, one just needs to understand SPARC. SPARC is not a single vendor, single designer - rather it is a specification that people build to. There are multiple vendors, multiple ideas, multiple implementations, and binary compatibility. There is competition in that ecosystem, with economics making the decisions over what will survive.
The "T" series was a design which came out of academia. A professor, Oyekunle Olukotun, figured out that a processor could provide more throughput and consume a lot less resources by leveraging highly threaded architecture. Of course, he was right.
http://en.wikipedia.org/wiki/Afara_Websystems
Afara was purchased by Sun Microsystems, and Sun brought the processor to market. The T1 performed integer and crypto operations many times faster, socket-per-socket, then existing competitors. It created waves in the industry.
Sun released the T2 saw 2x integer/crypto threads (and throughput), 8x floating point cores (and throughput) with a process shrink (and included a way-oversized memory bus that could never be fully utilized, because of advancements in memory technology that was serendipitously released.) Using the same process, the T2+ came out shortly afterwards, yanking 10GigE, yanking some under-utilized memory bandwidth circuitry, and adding SMP links for up to 4x sockets. One would suspect the T2 & T2+ were done in parallel since they were both done with the same process.
This progress Sun made with the T processors is very respectable, leveraging the same design principles, while remaining within design constraints that competing vendors (with far deeper pockets) did not have. Other vendors could afford to build fabs and shrink their process while Sun continued to leverage fabs with older process from TI and used advanced threading design to compete on throughput.
The T processor was never intended to be fast single-threaded performance, Rock was supposed to have that.
http://en.wikipedia.org/wiki/Rock_processor
There was a lot of speculation as to when RK would be released, whether it was canceled, but very little speculation as to why Rock never came to fruition. Looking at it in hindsight, I suspect the throughput of Sun T2, Sun RK, and Fujitsu VII on the same process (.065nm) was probably not far enough apart to justify RK's existence, especially when another process shrink was needed and the market estimations probably could not justify a business case for development of 3 processor families and their systems on a new process.
Intel was caught by surprise - it seems they abandoned their roadmap, cobbled together Multi-Chip Modules with hyper-threading to compete with T, and threw around money (i.e. bribe Dell, start new chip lines, etc.) to hold off the competition as they designed new processors in a new fab processes - a recent fab:
http://www.eetimes.com/electronics-news/4213295/Intel-to-build-new-Arizona-fab-
AMD sped ahead with 64bit x86 compatibility, but stumbled on a 4-core processor on a piece of silicon, in their attempt to scale threads up. People wondered if they would survive. They had to look for external investors to build new fabs. They could not compete with Intel without investors.
http://www.marketwatch.com/story/what-amds-fab-spinoff-means-for-intel-industry
To increase performance, a process shrink was really needed, Sun's partner, TI, did not appear to have deep enough pockets to do it alone - so it looks like they had to work together with TSMC.
http://business.highbeam.com/3094/article-1G1-175301052/tsmc-45nm-deal-sparc-sun-chip-goals
Looking at the T3 and T4, it appears they were both created with the same new process, probably done in parallel. The T3 doubled cores, experienced 2x integer/crypto/floating with the fab shrink, using the new partner (making Sun/Oracle able to come to market more quickly.) The T4 required a whole new core (which took longer to design and come to market by Oracle.) I suspect the core used in the T4 used lessons-learned from RK and those two processor lines basically consolidated with the shrink to the new process - a wise cost decision, but fumbling probably cost Sun their independence as a company.
This being said, with the T5 (2012?) and T{next} (2013?), one might expect a similar parallel strategy... shrink, double existing cores, and add another SMP bit for double sockets (phase 1), add advanced new core features (phase 2.)
Note in the T5 info released thus far, memory versioning - this looks a lot like Transactional Memory from RK to me. I suggested to people in Sun, when they released their storage systems based upon OpenSolaris, that they needed to bundle compression with their crypto units to speed ZFS - glad to see it announced. :-)
Sorry Larry, you're too greedy for my tastes. I'm in the process of moving off my old Sun iron onto x86 and off Oracle DBs (and Apps too.) I was in your camp till you shafted me with the HP BS then elected to only support your own Linux flavor, then only your own VM flavor - you know, I'm seeing a trend here.. How long before you announce you're not going to support Power?
The only folks likely to go for the T4 are those who can't move off SPARC for arcane reasons or those too stupid to avoid your taxes.
Now, notwithstanding that I have to be considered utterly biased due to my current terms of employment.
AIX on power vs HPUX on (either PArisc or Itanic) vs Linux on X86(_64) vs S10 on T4.
or god forbid M$ on X86.
Benchmarks as published by **any** hardware vendor on their OS, on their hardware, are going to be absolutely the best possible set of tweaks that can be applied to get the numbers they need to go to market.
OH!!!! Benchmarks are NOT technical results... .they're ... . OMG Marketing data!!!!!
That out of the way now?
Good.
Sitting where I am (Allison can likely look me up), I have this to say about operating systems, platforms and results.
Architecturally, every hardware platform and OS combination has its strengths, and every task has its specific load points. Choosing the correct hardware and OS combination to get the job at hand done and done well for the best ROI is an art form.
Willy wagging and benchmark barfing are NOT art forms. They are troll forms.
Oh --- someone said troll!!
Off to play warcraft, my ADD just kicked in.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
