Posts by Jesper Frimann 478 publicly visible posts • joined 8 Oct 2008
The key here is reading the small print.
It's a brand new Oracle system.. versus a POWER5 system.. which was GA 2004.......
IMHO this just shows how desperate Oracle is.. they need to find and pick a fight with solutions that are not just current generation -1 from their competitors.. but systems that are up to 8 years old.
// Jesper
I must admit I stand with TPM. What Oracle is claiming in pricing against HP and IBM is actually ... hilarious.
Now I have nothing against people saying.. I want to have a good comfortable chair, but I don't want to spend money on a quality crafted designer chair, that will last a lifetime, but I'll go get one from from a place like IKEA cause it's good enough for me.
But when they then go and buy a mass-produced industry standard chair from Snurakle that costs the same as the quality crafted designer chair, they need their heads examined.
Basically what they are doing is taking some ... well lets call it lowend midrange servers and storage, that have been put together into a clustered solution, which they have smacked a 'highend' sticker on.
A system that does not integrate with anything else. neither horizontal or vertical in the solution stack.
Furthermore it is just as proprietary and closed as an IBM 80'es mainframe.
And at last they put a serious price tag on it. A price tag of such a magnitude that it will only make remotely sense to clients if they (Oracle) try to compare it against REAL highend solutions from other vendors, and even HighEnd solutions that have been neutered and purposely made suboptimal solutions.
Again, if I were to compare a HP or IBM scale up solution (cause if you compare scale out versus scale out for the different vendors the machines are quite different) against a
For HP you are kind of stuck with the SD2, again if you were scale out I'd go with a HP itanium blade, but using 2 x SD2-16 is ridiculous, one is quite enough. And as others have stated then the storage system is.. well.. why do you need to compare a highend storage system (that isn't even sold anymore) , that is designed to have many many host attached to it, with a buildin proprietary system ?
For IBM again as it is with HP why go with a highend system ? And also the server chosen is a highend server, why use a highend POWER 780 when you really only need half a midrange POWER 770 ?
Now if you were to compare these realistic HP and IBM configurations against the Oracle solution, you'd see that both the HP and the IBM solutions have been cut by more than 56% of the price.
So what would you buy.. the lowend midrange eq. bundle from Oracle, or the high quality gear with all it's RAS from HP or IBM ?
// Jesper
@Phil 4.
You are presuming that a DCM module will have double the number of cores, compared to a SCM. It seems much more likely that a DCM module will house two six core chips, clocked a little lower than the the SCM modules.
And from a heat envelope perspective it also sounds a bit more realistic. I doubt that a process shrink, and some more dynamic power management will be enough to double the number of cores per socket, without additional cooling measures.
// Jesper
Your idea with building a cloud of POWER 775's is actually one that we, where I work examined, back in 2006, here we actually had the POWER 575 as our intended platform for our shared environment.
Today we use POWER 770's, the problem with using the HPC nodes for commercial workloads is the RAM per core numbers. The POWER 775 has 256 cores to 1 TB of memory that is 4GB of memory per core. This is way way to little for todays commercial applications, which is what UNIX systems run today.
Now for something like for example a SAP system and databases (specially if you use DB compression) you will normally have something between 8-16 GB of RAM per core + RAM for OS and virtualization.. Furthermore cause you want to drive up utilization on your platform you will normally overcommit your processor resources, usually somewhere from a factor of 2-4.
If you look at individual virtual machines that have been sized for peak workloads, their average utilization is usually somewhere in the 10-25% range.
So the actual physical amount of RAM you need per core is usually something like the factor you overcommit with times the non overcommitted amount of RAM per core. Hence usually somewhere between 2-4 times 8-16+ = 16-64 GB of RAM per core. So on a machine like the POWER 770-MMC with 64 cores you would need 64 cores times 16-64 GB of RAM + overhead. Or realistic something like 2-4TB of RAM. For a machine like the 775 this would be in the 8-16TB range.
So the HPC nodes aren't really suitable for commercial workloads these days IMHO..Sure there is memory compression but that won't give you enough IMHO.
Now as for the DCM modules I would guess that these would be 2x6 core modules not full 2x8 core modules, but lets see what is announced here what.. next week is it ?
Thanx for a good article btw.
// Jesper
Your tunnel vision of reality is actually sometimes disturbing. It's like you think that the IT industry is 5-6 years old, and you are constantly rewriting the real history of the IT-industry to fit into that tunnel vision.
"AIX WPAR is a Zones clone", are you completely clueless or ? The general term behind Wpar's and Zones is called Operating system level virtualization. And Keb. the most basic versions of operating system level-virtualization have been around since ... 1979.. (http://en.wikipedia.org/wiki/Operating_system-level_virtualization)
IMHO Solaris Zones was an almost panic SUN response to the IBM hypervisor based POWERVM virtualization, which IMHO the SPARC Solaris combo still haven't matched fully here 7 years after POWERVM was released.
"AIX ProbeVue is a DTrace clone"
Again the idea of consolidating several tracing tool into one package is nothing new, tools like gprof comes to mind, which I remember way back from my university days when I ran BSD 4.2/4.3 on Mini and Micro Vax'es. Up until Solaris got DTrace it was IMHO a good deal behind AIX on the trace front. Sure the AIX tools weren't consolidated nor particular userfriendly... but hey that's UNIX for you.
And Keb, the rest of your post about Niagara just exhibits your lack of understanding of Server Computing. The T4 based systems are still a niche player, that is best suited for running Web Style workloads, where it can utilize it's accelerators and huge amounts of threads.
Sorry, IMHO the SPARC64-X looks like a much better choice for SPARC customers than the T5.
// Jesper
You sound just like a salesguy from DELL. Are you trying to make yourself a Stewart Alsop legacy ?
For your education that was the guy who made this quote:
“I predict that the last mainframe will
be unplugged on March 15, 1996.”
– Stewart Alsop, March 1991
He was at the time editor-in-chief and executive vice-president of InfoWorld, at the time.
Now more than 21 years later the Mainframe business is still a multi billion USD business.
Your whole chain of argument is ridiculous, you forget that POWER is absolutely dominating the UNIX marked. Now IBM doesn't only have 50%+ markedshare in Q4's but also in other quarters see TPM's article here: http://www.theregister.co.uk/2012/09/04/gartner_q2_2012_server_numbers/
// Jesper
I must admit that it's kind of hilarious to read the comments here.
@Bill
You really should try to read a little about POWER before making such ridiculous claims. Now there is a reason why IBM managed to get 50% plus UNIX markedshare in Q2 with a product that is 2 1/2 years old. Will IBM's UNIX markedshare reach 60% this year?
It's actually sad.
@David Halko.
With regards to Dynamically switch between single and multithreaded mode.
You do know that there is feature in AIX/POWERVM/POWER6/7 called Processor folding right ? This feature have been available since 2007, a little time after the POWER6 launch. It folds virtual processors together so that if you only have X active threads on a X core 2/4*X threaded virtual machine then it'll only schedule one Logical processor to one Virtual processor to one physical processor. So... Sparc and Solaris is not in front.. they are, as they have been for almost the last decade... behind.
Glueless 8 socket.
Are you sure it's not you who are clueless (sorry couldn't resist it)? Please direct me to a paper describing the gluechips needed to make POWER systems scale to 8/16 or 32 sockets ?
With regards to POWER7+ being late.
Yes, it's late. But is that due to redesigns, problems or simply cause it's been absolutely killing the competition ? Basically it looks like it's going to be POWER that it the last man standing to pick a fight with x86, unless ARM decides to grow up :)=
With regards to performance.
Platforms like POWER, Itanium and SPARC64 have traditionally focused on different workloads compared to SPARC and x86. That is the real reason why you haven't seen POWER adding the type of accelerators, that you see it do now before now. Furthermore you see SPARC64 add Decimal Floating point now, rather than adding the encryption and random hardware. And I don't really know what Poulson is going to add if any.
Again where is the JBB2005, TPC-C/E, specint/fp benchmark numbers for the T4 ? There aren't any, they have only done benchmarks where either their software or their accelerators gave them and edge. Sad really...
@Matt Bryant.
With regards to mx2 it was a terrific HP engineering solution to the fact that dual core PARISC processors outperformed Itanium single core counterparts. It was a great solution, although expensive AFAIR :)=
// Jesper
The difference between being that POWER managed to keep increasing single core throughput.
POWER5+ at 2.2GHz 16 cores 8 chips 217 in specint_rate2006.
POWER6 at 4.7 GHz 16 cores 8 chips 484 in specint_rate2006
POWER6+ at 5.0 GHz 16 cores 8 chips 542 in specint_rate2006
POWER7 at 3.86 GHz 16 cores 2 chips 652 in specint_rate2006
So keb, lower Ghz doesn't always means slower cores.
// Jesper
From what I am hearing. It seems like EDS tactics is slaughter HPUX enterprise installations and put them on scale out x86 blades. When it then doesn't perform, then you throw more HW at it. Now having cleaned up a lot of server sprawl in my days I guess that is kind of a short sighted way of doing things.
But then again I also work for the EDS competition :)=
// Jesper
@af Horunge
I disagreed with your arguments and only the arguments, not your attack on the validity of "RISC just CAN enable better/more single-cycle performance" in the case of SB-EP versus POWER7. Basically POWER was never the pure RISC processor and Intel x86 processors aren't really a CICS processor, and haven't been for years.
And IMHO you are still wrong with many of your arguments.
"and how far their system interfaces really scale etc. is not relevant, as we are not comparing systems but how fast their cores run"
Sure it is, 'interfaces' take up chip real estate, adds to memory latency etc. etc. hence it has an impact on performance, also for single core throughput.
And if you look at Turbo boosts for both processors then running one core they will surely run at peak speed, both of them.
// Jesper
Eh, you can read a benchmark submission can you ?
"Intel Turbo Boost Technology up to 3.80 GHz" versus"Intelligent Energy Optimization enabled,
up to 3.94 GHz" for POWER7.
So the processors run at practically the same clock frequency.
With regards to design points.
Then sure it matters if the processor is designed for 32 sockets or 2 sockets. POWER7 chip has 3 Intra node busses, and 2 internode busses, compared to the 2xQPI links that the SB-EP uses to connect two processors to each other. Again scalability comes at a price.
With regards to an imaginary comparison of cores, then sure you can do that, but it will always be in the context of two multicore chips, that are designed for different types of servers and on different production technologies. If you really wanted to compare core performance on a theoretical levels as you are suggesting, then you basically would need to 1 core chips manufactured and designed for the same types of servers. A theoretical adventure.
Furthermore when you compare SB to POWER7 you have to take into account that one is sparkling new, and the other is a 2 1/2 year old product. Which kind of makes your theoretical comparison.. well shaky.
Furthermore when it comes to accelerators, then on a benchmark like 462.lipquantum, Xeon processor have managed to increase the per core score with a factor of 7 going from pre AES Xeons to the current SB-EP.
So where I might agree that SB-ep is a great processor, then I think that the trees don't grow into the heavens.
// Jesper
@af Horunge
You start to use SPEC® CFP/INT2006 Results to compare processors, now these benchmarks are supposed to tell you about single threaded performance, which haven't directly got much to do with 'processor performance'.
If you want to look at 'processor performance' or throughput you really need to look at the rate numbers.
Now if you think that Memory capacity and SDD disks have any what so ever impact on SPECint/fp®_rate2006, you are quite wrong.
As for accelerators, then for example POWER7 has Decimal floating point support which SB hasn't, and SB has AES support which POWER7 hasn't.
// Jesper
Eh ?
You do understand that SPECint2006 and specfp2006 (not rate) have been broken and has been for quite some years ?
You also do realise that POWER and Sandy Bridge-EP are targeted at completely different types size and types of systems as POWER has it's sweet spot at 8 sockets and .. well SB-EP at 1 ?
You also realise that the increased use of 'accelerators' inside processors today make it increasingly hard to make general direct comparisons ?
So I think using the phrase "If you didn't know that, I'm not sure what you're even doing on these forums.", is a bit harsh.
// Jesper
Well when you deal with IT vendors that supply software and hardware, they are always trying to get an as big as possible slice of your IT budget pie.
Hence if you as a software company can convince your customer to use low grade cheaper hardware, then there is more IT budget left to be spend on software. Hence you have a chance to get more revenue.
Personally I'd rather spend money on quality hardware than on additional software licenses that aren't really needed. Again the hardware will bring value, added software licenses won't.
// Jesper
Itanium ?
All the Oracle software sales reps I have encountered for years years been saying low socket count bare metal Wintel boxes, the cheaper the better. And then smack RAC ontop of that.
The reason is that they then get a bigger part of the IT budget pie.
// Jesper
Jens, I am pretty sure I've even seen you in RL. And doing what you do (or did) you should know that benchmarking is all about 5 things
Benchmarketing, Benchmarketing, Benchmarketing, sizing info and testing.
So surely people should speculate, that is what 80% of the people here do 99% of the time. Most of them without a clue.
Now surely you are absolutely right that doing benchmarks is a huge task that takes a lot of resources, but when you put your neck out, it being you guys, EMC, IBM, NetApp, HP, HDS whoever, then it's at your own risk, and people like Kebabbart will twist and turn any results you have put out there to fit his agenda.
Others will try to make sense of it trying to figure out what can be learned from it.
So if you don't want your benchmarks analysed, misinterpreted, admired, scoffed at whatever, then don't put it out there.
And remember that when you sign your posts with your company's name and your title. You are perceived as speaking for your company.
// Jesper
Kebabbert.
The story is here:
www.version2.dk/artikel/netgroup-kunde-rasende-over-nedbruds-forklaring-der-er-nogen-der-lyver-45482
And yes there are stories out there with more or less all types of systems and technology that have failed, and you are always sure to point this out, as long as this does not involve any Oracle stuff. So I thought I'd just help you a bit, so that you could get full circle.
// Jesper
So installing the storage-server software package (comstar) on a blade to present ZFS storage through iSCSI to client nodes is custom hacked ?
I thought you installed software, configured it and then used it on Solaris. Not custom-hacked it in.
And it was, according to what people have been able to dig up, a hardware failure of the comstar/ZFS node that trashed the lot, not the underlying HDS storage.
Now that is kind of disturbing, feel free to look at the news articles, they are in danish so you'll need to translate.
// Jesper
Yes, lets hear the ZFS is the best in the world story bla bla bla.
Try telling it to the clients of one of the larger cloud suppliers here i Denmark who build their cloud infrastructure around redundant Oracle Comstar servers with ZFS.
All the clients who didn't have backup.. lost their data.
Feel free to look up the story.
// Jesper
Come on Matt, the amount of people who buys x86 and really cares about the HW is tiny. If they did care about the HW they'd buy something else .)=
Now the management interfaces and surveillance software stack that is another thing.
And it still doesn't change that.. with regards to Itanium.. then it's ... it's dead Jim. Not that I gain any pleasure from seeing another Unix platform being put on deathwatch.
// Jesper
Come on Matt.
HP involvement on Linux on Itanium is kind of non existent today.
When I look in my bookmarks under Itanium, I find that
http://www.itaniumsolutionsalliance.org/ gives me server not found.
http://www.gelato.org/ has no activity on it..
http://h20341.www2.hp.com/integrity/w1/en/os/linux-on-integrity-overview.html has an alarming amount of references to Linux on Proliant.
So IMHO it's not like that expertise is there anymore.....
And when it comes to the linux Kernel HP isn't even in the top 20 of contributors.
Not that HP couldn't do a their own Linux distro. But they would have to build it up from the ground, unless they were to go and snatch up a company like Redhat, but personally I don't think that they want to shell out 10-15 Billions right now, for a company like that.
// jesper
No. I haven't seen the SPARC roadmap you've linked to.
But I guess it's this one:
http://www.oracle.com/us/products/servers-storage/servers/sparc-enterprise/public-sparc-roadmap-421264.pdf
It's not a SPARC roadmap. It's a SPARC based SERVER roadmap. That is a vastly different thing.
What we are going to see next year for SPARC is a shrink from 40nm -> 28nm, and an enabling of the T4+, or T5 as they might call it, to go to 8 sockets, while being in a T-series. Then the ageing Mseries is going to get rid of the obsolete SPARC64 and replace it with the T4+/T5.
Now T series going for 2,5 in throughput while doubling the number of sockets is bold, and I have my doubts there, unless new accelerators are introduced to lift throughput on certain benchmarks.
On MSeries, it's going to be a cold day in hell before replacing the SPARC64 VII+ with a T5 will give you a boost of x6 on all benchmarks.
They Will IMHO be able to pull it off on TPC-H where the SPARC T4 results are stunningly good, but on something like SPECINT or general purpose industry standard benchmarks.. Forget it.
On something like TPC-H the T4 is aprox x4 faster per core than SPARC64 VII+, where on something like SPECINT (judging form the T3 results) they are more or less equal.
So sorry but you are eating the marketing cake a bit to happily after my taste and not really trying to understand what is being said.
// Jesper
Now this is a processor roadmap:
http://regmedia.co.uk/2009/09/11/sun_sparc_roadmap.jpg
And as you can see Oracle is basically just delivering on what SUN started, there really haven't been released any new info since this roadmap on the processor side.
@Kebabbert
I think you are being way way to optimistic, on Oracle's behalf.
I mean if you look at the TX processor based line of servers, then going from T1 based servers to T2 based servers was a huge improvement. You saw an improvement of almost a factor of 4 going from T1 to T2, both on a per core and a per chip throughput, on most benchmarks.
Going from T2 -> T2+ .. well nothing on a per chip level, and the same from a per core level, actually the throughput decreased a little.
Now going from T2+ -> T3 again cost on the per core level, but you gained almost a factor of 2 on the per chip level.
Going from T3 -> T4 you gained almost nothing on the per chip level but a huge boost on the per core level.
So sure you had a huge improvement from T1-> T2 but from there on you've only managed to get around a factor of 2 on per chip throughput in general compute power going from T2 over T2+ over T3 to the T4. And that is in 5 years. So sorry the history of the T SPARC's does not imply a game changing marked winner of a processor for the future.
So it might be that Oracle is able to turn the Tx Sparc architecture into a processor that runs Oracles own software really really great with various accelerators that is tailored at their own products. But again the business model of IBM in the 70ies and 80ies is a risky one, it almost killed off IBM.
And to be honest not many clients today like Oracle.... again it's like people felt about IBM back in the 80ies and 70ties. But you are most likely to young to remember that.
// jesper
@King1Con
I'm glad that this works for you. But the problem for Oracle is that where as they can replace servers with higher socket count, that appeared 7-8 years ago with servers that have 1/4th of the number of sockets. The competition are doing replacement of their own servers with servers that have 1/8th of the number of sockets.
And back then, Oracles competition was already faster than them by almost a factor of 2.
So where as T4 have helped greatly on their competitive situation , and is a great product, the competition have managed to pull further away.
// Jesper
Well IMHO and as TPM also described in this article http://www.theregister.co.uk/2011/10/14/ibm_power7_plus_redux/, POWER7+ is 'late'.
When I write 'late' I don't mean late as in what you Keb would like it to be, problems.. bugs etc.
see for example Charlie D's article here: http://semiaccurate.com/2012/03/21/ibm-power-7-spotted-and-it-is-a-monster/ , but 'late' as in they didn't need to release a new Processor.
If sales are strong and the competition isn't really releasing anything that will affect you, why release a new product ?
Again the Q4 launch of the 'C' generation of machines was perhaps not a big one to the untrained eye. I mean 200MHz more on the processors on the POWER 770 ? What's the big deal?
But there were more under the covers, the whole Internal IO system got a much needed upgrade, doubling the amount of bandwidth and number of IO transactions that the 'internal' IO system is capable off. Furthermore the amount of memory supported doubled.
And there were also some other smaller things.
And with regards to the 'new' Linux only servers. Well you obvious haven't been following the UNIX marked for a very long time have you ?
I mean this is not the first time that IBM have tried this approach, they did it with the OpenPower servers back in 2005. So again nothing new.
// Jesper
@AC
"Finally, there are ISV licensing costs on Power7 vs. x86. Don't even think about running Oracle software such as their database or WebLogic on these machines."
Well it depends, it's not a clear cut case, as you try to portrait it.
If you use per user license, then your argument is not valid.
If you are running on only 1 core.. then the price is the same.
And then there's the whole virtualization angle, where for example running Oracle software on VMware is a nogo, but on POWER it's more or less the default.
But if you are your normal average shop, that runs a few Oracle servers on bare metal where you pay for processor licenses, sure.. then it's normally more expensive to use POWER rather than x86.
// jesper
@AC
.....
"Anyone remember how that worked out for them? IBM have rather bigger reserves and better diversity."
Well the main difference I remember, between DEC and the other *NIX vendors, was that the DEC sales people had tailor made suits with gold cufflinks, just like the IBM mainframe sales people.
Now the HP/SUN and IBM *NIX sales people didn't.
// Jesper
"IBM trying to put Power up against Linux on x64 is an instant fail. Hard luck, mainframers!"
There is no doubt that with current volumes, then the traditional UNIX hardware cannot compete with x86 when it comes to TCA for the Hardware.
But when you start to put virtualization layer, takes utilization into account etc etc. then things even out. On our local 'cloud like' offering, it is actually cheaper to get a Virtual AIX server than a Virtual Windows server of equal capacity.
And it's cheap shit 2 socket x86 versus 8 socket 4TB RAM POWER servers.
// Jesper
Actually I think there is kind of an underlying movement in the market. A consolidation and movement towards that suppliers can do an (almost) vertical solution stack. This is kind of like back to the future, to a time where you got your whole IT from a single vendor. This is IMHO much more a trend rather than partnerships between different vendors in the solution stack.
I don't think this is a good thing for us that have to procure the whole software and hardware stacks. I think we are going to see less open standards, less portability and more vendor lock-in.
If this movement continues, and it is a big IF, there are going to be further consolidations, and to be quite frank then HP wouldn't be one of the companies that would be able to buy up other big companies quickly right now. I mean HP's long term debt is more than 50% of the current marked cap of the company. I know that the debt is in practice deducted from the maket cap, but it's still a huge chunk of depth. For comparison IBM (although having more long term debt than HP) it's still less than 15% of the total company cap, around the same % as Cisco and Oracle has.
So again if, and there is a lot of if's here, this vertical trend continues, then personally I think HP needs to merge with someone.
// Jesper
Well, why talk about benchmarks that supposedly exist in HP labs. Why not simply look at the ones that are out there.
From what I've been able to see then the p460 will outperform the BL870c i2 by a factor of.. well four at aprox the same price. It'll outperform the BL890c i2 with a factor of two at aprox half the price.
Sorry but Tukwila is capable of competing with POWER6(+) but against POWER7 and current Xeons' it not even close. I mean it even get creamed by SPARC.
And last.. then people buy hardware to run software on, and the club of software that runs ontop of Itanium is shrinking fast.
I take no pleasure in stating that IMHO Itanium is a dead man walking.
// Jesper
Keb... that story is now... more than 9 years old.
Next you'll claim that IBM will stop selling computers cause, they believe that we only need one machine per continent.... based upon an old quote by a IBM top exec.
You are simply making a fool of yourself, I am sorry, there is really no other way of expressing it.
POWER/AIX is most likely a more strategic platform for IBM than their mainframe business is.
It most likely generates more overall revenue for IBM than Mainframe does.
Why start til talk about POWER8 when we obvious haven't even seen POWER7+ yet. According to sources on the net it'll feature up to 4 chips per socket, which will give POWER servers the ability to .. more or less annihilate the competition.
And before you dismiss POWER7+, then remember that last time that IBM did a MCM based system for their 'ordinary' machines, was on the p560Q, back in 2006 which actually held it's grounds up until POWER7 was introduced. It's seldom that you see a system that holds it's grounds in so many years.
// Jesper
Kebbabert.....
"Do you deny my IBM links, where IBM executives talk about killing off AIX? Those links are not crap nor BS. I am not making up things, that you like to do. I post credible links. Links that go to IBM executives must be considered more credible than your made up FUD with no links."
That exec i Steve Mills, he is actually the president of IBM's hardware and software division, if he really wanted to kill off AIX, why haven't he done it ?
Ohhh.. perhaps cause IBM last year had a ... 46% of the UNIX marked. And a stunning 53% in Q4.
Sooooo..yes... the evidence is there.. IBM is getting out of the UNIX marked. *CACKLE*
"Regarding your details, you are wrong as usual. For instance, if you want POWER7 to increase performance, you can switch on some feature (single thread at higher clock speed? or something similar), but to do that, you need to reboot the POWER7 server. So there is no choosing "on the fly". :o)"
Perhaps you need to try just to kind of research what you are talking about.. you are talking about Turbocore mode, where you disable half the Processor cores on a chip and then run the remaining at a higher clock, with more cache per core.
What Alison is talking about is threading....
// Jesper
>>This is merely a way to sell the procs with broken cores.
>>It is not a feature. You'd think that since IBM owns the fab they could do a better job.
Actually IMHO you are both right. Sure less enabled cores is a way to use the chips with fewer working cores. But as many cores a possible in a machine isn't always the goal. Our standard POWER machine, is a 48 core 2TB 770-MMB. Simply cause we have found that for our usage, 64 POWER7 cores for 2TB RAM are to many cores. So again IMHO you are both right.
>>Very misleading. IBM shops must change the thread count based upon the load,
>>and the system does not adjust to the load in a dynamic way (Like SPARC T4 does).
>>This means you cannot have a mixed workload in one instance.
That is not correct BILL, in fact you are 100% off the mark here IMHO.
AIX and the Hypervisor actually utilizes something called 'processor folding' where virtual processors are folded together, hence in a kind of standby state, when they aren't needed. If you have ever watched a monitoring tool of a AIX machine with processor folding enabled, you'll see AIX and the Hyperviser acting together and compressing the load together on fewer processors. Hence it's far less static than other processors.
IMHO AIX+POWERVM+POWER are to be quite honest years in front of Solaris+SPARC. You have to remember where the T4 is the first 'Simultanious multithreading' enabled SPARC processor, IBM have been shipping processors with their SMT version for 8 years now. And it is tightly integrated into both AIX and the virtualization layer. Come on.. you know I'm right if you think about it.
>>Oracle has 16 threads running simultaneously on T3 and 8 Running Simultaneously on T4.
>>The "switching" you speak of is to handle cache misses on each core
>>(which believe it or not, even IBM has a lot of!)
>>I think even Kebabart could explain this to you if you are confused.
You are wrong, with regards to simultanious on the T3. It is what is known as fine grained multithreading, but the T4 does AFAIK use simultaneous multithreading. I must say I haven't tried out the T4 yet, although when I wrote our standard buying catalogue. I put the T4 in as the SPARC standard machine, that we have to use. I hope we can put off investing in larger SPARC's until the roadmap for those is more clear. I don't wanna do forklift upgrades if I don't have to.
Now I have to admit that I personally didn't like the T1-T2-T3, concept but T4, although not in the same league as Xeons and POWER when it comes to throughput, is a pretty good CPU.
>>I do understand a little why Oracle not reselling SPARC64 confuses you,
>>as you come from IBM, where rewrites/recompiles of code between different
>>versions of SW/HW is common, but Oracle/Sun have kept binary compatibility
>>for decades. They've promised to do this moving forward as well.
Sorry I think it's you who is fudding here BILL. Both Solris and AIX are IMHO 'the good guys' when it comes to binary compatibility, and following the UNIX standards. But the difference between T[1-3] and the rest of the SPARC processors are huge. Not in binary compatibility, but in how you have to use them. And I know for a fact that we have advised several clients against using T[1-3] based hardware and put them on Mseries up to now, cause it was cheaper to buy more expensive Mseries iron, rather than retuning their applications.
>>Of course, since IBM does not have an entereprise OS they can put on their comparable systems..
Both Solaris and AIX are enterprise OS'es, claiming anything else is .. well.. not serious.
>>I'm really impressed with the rate of innovation in SPARC since Oracle took over.
>>IBM should be very concerned. I don't think they will be able to keep up in the medium
>>to long run at this rate of change. Of course, Oracle has kept all of their SPARC promises so far, but it will be interesting to see if they can keep them moving forward.
I don't think you are right. First of alle Oracle is getting hammered, try to google for some marked share numbers, like this link (http://www.idc.com/getdoc.jsp?containerId=prUS22998411). Sure their business is growing, but it's way way below the marked. So come on...
Larry is turning the Oracle+SUN businesses into pre Gershner IBM, a business that only sells you a complete stack of their own stuff, at a price that is set not as a % profit, but rather as what they can get away with charging you. The percentage of Oracle clients that hate Oracle is increasing rapidly, and Oracle is increasingly standing alone. IBM HW sales rep will now bend backwards to push DB2 or Sybase(sap), where they used to don't care about what you ran on their kit. And HP is in an outright war with Oracle.
And as a guy who soon has his been with UNIX for 25 years, and who grew up on Solaris on SPARC as one of 2 platforms, I am sad seeing Larry perverting SUN.
// Jesper
Do you have some kind of special cut and paste buffer for this rant ?
It's the same and the same and the same and the same and the same and the same again and again.. and it doesn't really change over the years.
// Jesper (beer cause it's friday)
Well, if they are replacing SUN equipment from 2005, with POWER 780's, then we are most likely talking E25K machines and E6900 . Now they will most certainly also have been able to utilize these newer machines better than the old, so that the actual RL capacity might be anywhere from 50%-200% higher. And if installed capacity is only 16% more then we might be talking capacity that corresponds to 20-40 E25K's depending on how filled up the machines are/were.
Which kind of means that the savings in artificial Datacenter cost, would most likely pay for the new hardware and then some over the next 5 years.
Having old hardware standing around beyond it's expected lifetime is expensive.
// Jesper
Well most of the electricity prices are vat and other tax related stuff. At least here in Denmark it is.
I think one KWH cost around 28 euro cent, here and of that only 4-7 euro cent is actually the cost of the electricity.
In Denmark windmill electricity is subsidised, so that the companies that build and run wind farms get a minimum electricity price for a fixed number of years, which means that if the marked electricity prices are high they get nothing and if they are low they perhaps get 1-2 Euro cent per KWH. Again for us consumers it's peanuts compared to the total price of the electricity.
When we moved into the house we live in now, we had a bill of 3.100 Euro on Oil (old boiler), and 1600 Euro in electricity.
I've changed the boiler to a sparkling new natural gas condensing boiler and have changed the fridge etc. to energy efficient, stuff. Also I have changed most of the light bulbs, to LED where I can. The newest generation of LED bulbs are actually good for use in the house.
Now our total energy bill is around 2.700 euro a year, for an investment of 7.500 euro, and 3 years after with raising energy prices.. the investment have paid for it self.
Next step is to change the old windows and have the insulation overhauled.
// Jesper
Actually, some of the farms we have here in Denmark, located on the danish west coast, the efficiency is actually 40%+.. I've read a study that suggest 45% efficiency there, and from that you have to substract maintenance etc. I think that Horns Rev 1, which is owned by the swedish energy company Wattenfall, they have 42% realised efficiency.
So it's more placement than offshore, that makes the difference :)=
// Jesper
It gets even funnier it stands for Dansk Olie og NaturGas...... Danish Oil and Natural Gas in english, kind of like the Danish version of Gazprom.
DONG is actually all the old small local danish public power companies that have been bought up by one of companies. State owns 75% and a single commune 25%.
But they most likely won't be doing to many windmill farms anymore. The chairmain of the boards of directors just fired the whole top management due to the fact that the Director of the company, had given his top directors 'private company' level salaries.
The small group of execs that were fired where the ones that developed and executed the financing models and project office that E.ON is trying to copy, that allowed DONG to build windmill farms at x2 the speed of other players, and raising enough finance by making a unique model that allowed eg. pension fonds to invest in their projects.
This is actually quite a scandal right now in Denmark. Cause DONG is making huge profits to the Danish state coffers, and is actually extremely aggressive in pushing green energy. The chairman of the Board is also on the 'board' of the main 'conservative' party in Denmark, who has always hated Green energy, as they get most of their campain money from Coal and Gas companies... so.. the feeling here is kind of that the next Danish Big Danish Company who is on the verge of taking a leap to being a big Multinational company just got axed, and that their 'Green' business model is going to be axed.... not cause it isn't working.. but cause it does not fit with certain politicians, that have managed to move into the board of that company
// Jesper
The Philosopher's Song (Monty Python)
Immanuel Kant was a real pissant
Who was very rarely stable. Heidegger,
Heidegger was a boozy beggar
Who could think you under the table.
David Hume could out-consume Wilhelm Freidrich Hegel,
And Wittgenstein was a beery swine
Who was just as schloshed as Schlegel.
There's nothing Nietzsche couldn't teach ya'
http://www.metrolyrics.com/the-philosophers-song-lyrics-monty-python.html
'Bout the raising of the wrist. SOCRATES, HIMSELF, WAS PERMANENTLY PISSED...
John Stuart Mill, of his own free will,
On half a pint of shandy was particularly ill.
Plato, they say, could stick it away;
Half a crate of whiskey every day.
Aristotle, Aristotle was a bugger for the bottle,
Hobbes was fond of his dram,
And Rene Descartes was a drunken fart:
"I drink, therefore I am"
Yes, Socrates, himself, is particularly missed;
A lovely little thinker but a bugger when he's pissed!
// Jesper drinking homebrewed beer.
It's still the responsibility of the people who designed the solution.
If you make a solution using entry level components and/or midrange components and you promise highend features, availability etc. you can't blame it on the hardware vendor. At least not totally.
You have to know and understand the building blocks you use.
// Jesper
Well one thing one have to take into account when looking at POWER sales is the big jump in capacity going from the old pSeries platform you are upgrading to POWER7.
1 Socket machines
POWER5 -> 12,5 rPerf
POWER5Q -> 20,3 rPerf
POWER6 -> 14,7 rPerf
POWER7 -> 92,0 rPerf
2 Socket machines
POWER5 -> 24,9 rPerf
POWER5Q -> 38,3 rPerf
POWER6 -> 39,7 rPerf
POWER7 -> 176,6 rPerf
4 Socket machines
POWER5 -> n/a rPerf
POWER5Q -> 75,6 rPerf
POWER6 -> 78,6 rPerf
POWER7 -> 335 rPerf
8 Socket machines
POWER5 -> 96 rPerf
POWER5Q -> n/a rPerf
POWER6 -> 141,2 rPerf
POWER7 -> 692,5 rPerf
32 Socket machines
POWER5 ->393,6 rPerf
POWER5Q -> n/a rPerf
POWER6 -> 553 rPerf
POWER7 -> 2812,7 rPerf
Which basically means that when you are upgrading a Big POWER5 based p595 which costed you a fortume, had hundreds of adapters, and perhaps ran at 20-40% utilization you are now putting in for example a POWER 770 which cost 20% of the p595, runs at 40-70% utilization and use 16-24 adapters.
Furthermore the 770 is perhaps able to house the load of two p595.
So the fact that POWER revenue for IBM is increasing, means IMHO that a lot of capacity is being shipped, cause again you only have to spend around 10-25% on hardware to get the same capacity.
// Jesper
Well, iSeries does not, AFAIK, tricker a lot of POWER hardware sales today. When I look at the AS400 stuff we have still standing around from the premerge of p and i, then I can see that most of the solution is actually internal disks and tape drives etc. and then a little iSeries box.
So I guess that the majority of iSeries revenue for IBM have turned into revenue for their storage division, and only a little part have made it into the POWER 'division'.
// Jesper
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
