Everything Oracle does is about money. You have to look at any of its actions - no matter how peculiar or provocative they may be - through an economic lens, focusing on the benefit those actions will provide to Larry Ellison and the company he co-founded decades ago. Take last week's out-of-nowhere kick in the chips Ellison and …
All went wrong at the end there, didn't it?
"... is for HP to say to hell with Oracle....". Great, how?
"... to buy Red Hat for its operating system....". What? The same Red Hat that they already offer and that ceased support for Itanium last year? Oops.
"...EnterpriseDB for its Oracle-compatible implementation of PostgreSQL." There's a world of difference between "Oracle-compatible" and "Oracle". I don't fancy porting, recompiling, regression testing and fixing umpty-something thousand lines of code any more than the next man and that's before we start on the skills, training and support issues. This one also completely overlooks a Big Fat Hairy Deal, which is that Oracle do more than a database.....
HP are in the shit here and it's deep, smelly stuff too. This doesn't make too much difference short to medium term for the existing HP shops, expansions will still happen, upgrades will still take place and servers will be replaced. What it does do is leave them SOL when touting for new business and especially in trying to take business from competitors. You try touting for a few thousand high-end machines in a Data Centre when your Oracle roadmap has a "No Entry" sign visible around the corner.
RE: All went wrong at the end there, didn't it?
Have to agree, the idea of buying Red Hat seems a bit far fetched, especially as the market valuation of RH is rediculously high. Given Leo's background, I'd say he's much more likely to get cuddly or even acquisitive with SAP, which would give hp Sybase to fight back with and also mean they could cut Oracle out of the SAP supported stack. Now that would make Larry howl! I suppose it depends on how much Larry's "support license" blackmail fee is
I also don't see how lack of Oracle for Windows on Itanium was a problem as every Windows on Itanium jaunt I heard of used M$ SQL. All we did with it was round M$ SQL consolidation, which must have hurt M$ licensing-wise as we would consolidate literally hundreds of individual Xeon SQL servers onto a quite small Itanium cluster.
I am guessing the stuff on M$ SQL is not mission critical. If it is your upgrade path is FUBAR. Keep saying how it takes so many fewer Itanium than Xeons and how much better Itanium RAS features are. You might believe it but the market place doesn't. I am betting the vast majority of Itanium customers at this point were captured by lock in from the DEC legacy world.
"I am guessing the stuff on M$ SQL is not mission critical....." True, most of it was departmental stuff.
".....If it is your upgrade path is FUBAR...." Not really, we've left it all running for now whilst we look at the alternatives as most of it doesn't need any upgrading. When we need to upgrade to a new version of SQL Server to meet a business requirement then it will go onto x64, hopefully on large servers like DL980s.
"....I am betting the vast majority of Itanium customers at this point were captured by lock in from the DEC legacy world." The vast majority of hp's Integrity customers are using hp-ux, which had nothing to do with DEC, Compaq or anyone other than hp.
Sauce for the goose...
Not a bad article overall TPM, but as usual, you can't resist a bit of Itanium bashing whilst not applying similar "rules" to the other processors:
>> throwing out the "Kittson" name is no longer sufficient.
Ahh.. but throwing out the "Power 8" name is? What about the SPARC roadmap, that's even more vague?
Actually I think you will find that all the processor roadmaps out there extend roughly out to about 2015 - even Xeon's roadmap goes no further than that. So why do Intel need to deliver a longer roadmap for Itanium than everyone else?
If _anyone_ out there should be questioned about their processor roadmap, surely it should be SPARC, which already lags furthest behind in most reasonable asessments of performance, where it's only OS is already available on Xeon, and where the focus of its sales force is elsewhere (Exadata).
SPARC roadmap is clearer than, say, POWER. It is available, just look at it.
Regarding SPARC performance, yes, it has several world records if you just cared to look up. Have you heard about TPC-C? IBM has 10 million. HP has 4 million. Oracle has... 30 million. There are other world records as well, that Niagara T3 owns, just look at the wikipedia article. For TPC-H, Oracle is also beating IBM
It has a Disk size to database ration of 102.6 which is almost a factor of four worse than the last M9000 benchmark, which was also far out with a factor of 26.
Talk about stroking the disks. All the other submissions are no where near that factor. It's so large and obvious that it's actually ridiculous.
re: Sauce for the goose...
Pretty funny... TPM may be biased toward Power, but he has not been unfair to Itanic. Oracle has the most detailed roadmap out there. Power and Intel just show names... SPARC shows cores, estimated perf, and sockets supported. You can attack IBM on this fact, but Oracle has been very open.
Kebabbert - Stop!!
Will you please STOP will all the pointless benchmark quotes, it's all you ever bloody go on about your posts are utter drivel!
Seriously...are you even a sys admin?
Re re:Sauce for the goose...
Yes SPARC has been good to have a detailed roadmap, so has Itanium and POWER is always kind of sketchy.
Itanium processors are always late compared to the roadmap, POWER sometimes are late but as the roadmap isn't to detailed, it is sometimes hard to tell. But one thing is certain about POWER and Itanium is,THEY ACTUALLY BECOME PRODUCTS.
SPARC processors on the other hand are announced and then cancelled.
Lets see.. 2004 SPARC V (Millenium) Anybody remember that one ?
An what about 2009 SPARCK RK (Rock) Anybody remember that one ?
And the performance estimates are so far out and open for interpretation that you gotta be kidding..
We have been burned so many times by SUN roadmaps where I work the last 10 years, that there aren't many fans of SPARC left.
Not such a fare comparison....
> IBM has 10 million. HP has 4 million. Oracle has... 30 million.
The HP SD2 is a single machine i.e. non clustered running a single database while the Sun machines are clustered running RAC.
@AC 08:12 GMT
>>> throwing out the "Kittson" name is no longer sufficient.
>Ahh.. but throwing out the "Power 8" name is? What about the SPARC roadmap,
>that's even more vague?
Well, if IBM doesn't deliver on Power8, it'll be shooting itself in the foot.
If Su^H^HOracle doesn't deliver on Sparc, it'll be shooting itself in the foot.
If Intel doesn't deliver on Itanium, it'll be HP that takes the bullets. There are two OSes that now only run on Itanium: OpenVMS and HP-UX. Both are HP's.
SPARC64 is end of life
There will be no future chip. Fujitsu is out of the SPARC market. period end of discussion.
The "chip roadmap" that Oracle has out there is not a chip roadmap it is a "systems roadmap" which deceptively does not saw which chip is being used to hide the end of SPARC64.
Yes Oracle has 30million with 27 servers....hmm do the math on that one and calculate 27*4*16*.25 = 432 licenses * ($47.5K EE + 23K RAC + 11.5K compression + 11.5K partitioning)
and the grand total is $40 MILLION for just the Oracle software
Maybe HP should drop support for SPARC64 since we all know Fujitsu has plans to cancel it.
re: SPARC64 is end of life
Whether this is true or not, I don't think it matters. The roadmap shows a SPARC CPU taking over performance-wise (throughput AND single thread perf). This whole SPARC64 is dead thing is just a red herring. Who cares? SPARC is an open technology and Solaris will run on SPARC regardless of who makes it -- no recompiling! You don't get the same thing with Itanic nor Power -- they are truly propriety. Sun has used third party SPARC way before SPARC64 came around and Oracle may well use it in the future. Get some new FUD!
So? What is your point? That Oracle's benchmark score is not valid, that IBM still is faster because IBM "has faster cores"? Or are you critizicing the Oracle hardware? But why did you not criticize IBM's hardware earlier?
"....Almost 4 years ago IBM published a result of the p5-595. This system yielded with 1.9 Ghz Power5 CPU 100,512.30 QphH. 4 years of development, 256 GByte more memory and 2.6 times the frequency gives you just roundabout 50% more performance. Nice, but not that impressive, especially given the effort put into this CPU in regard of cycles...."
But no, why would you? When Oracle does something, you are quick to criticize and when IBM does the same thing - it is ok. I remember when I showed you a benchmark, and you said something like "so what about that low latency? IBM has higher throughput! Moahahaha!". Later I showed you another benchmark where Oracle had higher throughput, and you said something like "so what about throughput? IBM has lower latency! Moahahaha!".
Now you are doing the same thing. Again. With such a logic of yours, how can Oracle win?
I am sorry it bothers you, when I all the time post links and white papers and articles - to backup my claims. But I need to do that, because I would not want to say untrue things or things that are not verifiable?
I hope you do not prefer if people say things without backup or links that support their claims? That disturbs me. I mean, if I say "IBM executives has officially said that AIX will die and be replaced with Linux" then it is not a claim that I made up. There are substance in what I claim. I prefer to post verifiable things. So please bear with me, and always posting links to back up my claims.
Yes, I know Oracle has a clustered result. So what? Who is fastest? Who do you speak to, if you need the highest possible performance out there? Who is your daddy?
I agree Oracle uses a cluster. I agree Oracle is cheaper than IBM if we look at dollar/transaction. But who has the highest performance on the planet?
"...There will be no future chip. Fujitsu is out of the SPARC market. period end of discussion..."
Can you back that up with links or is it more FUD? Here is proof that there are more SPARC64 chips from Fujitsu
In fact, it seems that as long as no one posts a link to back up his claims, we should consider it as FUD. It would help a lot, if everyone posted links so we can see what is FUD and not FUD. FUD can not be backed up with credible links. But I dont FUD, I post links. That is the reason SplitBrain complains that I always post links to official benchmarks. Sorry, but I will continue to post links - I do not want to be accused of FUDing.
Keb... it is my job where I work now to look through benchmark results and sizing data to make sure that the solutions we deploy on SPARC,Itanium,POWER and to some degree x86 do hold water.
Hence I have to look though a lot of bull from all the vendors and decide what is right and what is wrong.. So I analyse and analyse until my face turns green.
And I have to admit when I do post on the internet I can't help but analysing there also.
And your link to c0t0d0s0 that I should have a look at that. Sure I've looked at that benchmark before, and if you just look at the score it seems that upgrading from a POWER5 based p595 to a POWER6 based POWER 595 was a bad idea. And you also seem to forget that there actually is a POWER7 system there. But hey you don't really want to compare to that now would you.
But again analysing the disclosure documents might tell another story.
Now first you have to understand the TPC-H benchmark, it's very storage intensive and actually pretty close to being embarrassingly parallel.
Now lets have a look at the recent POWER benchmarks in this category.
QphH Machine Cores disk Processor Adapters DB DB Load time OS
100512,3 p595 64 1152 POWER5 96 Oracle 4H AIX5
156537,3 POWER 595 64 288 POWER6 24 Sybase 10H AIX6
164747,2 POWER 780 32 52* POWER7 12 Sybase 1H Linux
And as you can see the POWER 595 kind of isn't up to the same speed as the other two benchmarks, But you can see that the POWER7 system rocks (it also uses SDD's) So it able to do some serious IO, which is also why it at 32 cores outperform's a POWER 595 with only half the cores.
An if we try to look at some key numbers here:
Machine QphH/disk QphH/core Diske space/DB size $/QphH QphH/rperf
p595 87 1571 12,9 53,0 328
POWER 595 544 2446 6,6 20,6 283
POWER 780 3168 5148 4,0 6,9 387
We can also see that the POWER 595 is not really up to the speed of the two others, the Power 595 which uses a shitload of disks, and the POWER 780 which uses a very fast IO system.
Now the POWER 780 result is kind of a curious one, it uses Sybase and LINUX on POWER, and you would expect a DB2/Oracle on AIX would have performed a good deal better. So again this is not the full potential of the hardware.
Note btw that the current M9000 benchmark in the same category, can't even match the old old POWER5 p595 in per core performance, and that is using a disk space/DB size of a factor of 100 and using no less than 128 8Gbit ports. So yes.. hmm.. Come on give us your 'It holds several world records' speach :)=
"Now you are doing the same thing. Again. With such a logic of yours, how can Oracle win?"
By making the best products.
And by not writing lease for 3 years prices in every benchmark they do, cause their software is to expensive.
Quite frankly I'd like to see Intel tell Oracle to pound sand and refuse to sell them Xeon chips. I think Oracle is getting more than a little too big for its britches. Given the lackluster server sales by Sun, I would bet it hurts Oracle way more than it hurts Intel.
Jup, Oracle is kind of turning into the much hated IBM of the 70ties and 80ties.
Lets just say that nobody I am talking to in the business, apart from people who actually work at Oracle, have high thoughts about that company right now.
Red Hat + HP = bad
HP buying Red Hat would be the death of Red Hat as we know it. Red Hat Linux thrives in part because, like Microsoft, it's a hardware vendor neutral platform. That's a big part of its appeal; an insurance policy against the machinations and/or failure of any one of the big hardware vendors. If it were owned by HP, IBM and Dell would cease certifying their hardware for Red Hat; sales would drop off and a significant slice of their subscription revenue would disappear along with it. No one who uses Red Hat - except for a few journalist rooting for the next story - want this to happen; certainly not Red Hat customers, employees and share holders. It would also send a very negative message to HP-UX customers and send ripples of confusion through HP's sales force, who've had it drilled into them to push HP-UX and Itanium over competing solutions at every opportunity.
It is not that hard to move DB
The enemy of my enemy is my friend?
The enemy (IBM) of my enemy (Oracle) is my friend?
IIRC, IMB's DB2 runs on AIX and Linux.
AIX and Tru64 were both succesors to OSF/1 which IBM and DEC colaborated on. Although Tru64 is no more, I'm sure IBM would be willing to build some bridges back to DEC's customers. Hey, with UNIX portability features built into OpenVMS, it shouldn't be a big deal to port.
The comment about HP getting its chequebook out is nothing new.. in times of old hardware vendors would contribute to the porting and maintenance costs of getting software onto their platforms.. my guess is that Oracle have asked for a much bigger wedge than before because the economics have changed.. It is only ultra high throughput scenarios where Itanium performance is justified, but that is precisely the niche that Oracle is targeting with ExaData2 (v2 replacing HP kit with Sun).
If Oracle is motivated by nixing a HP TPC-c benchmark, that starts to look like market abuse..
Re: Market abuse
I have to agree. If Oracle has as much of the database market as this article suggests, and Oracle are willing to 'talk terms' with HP... then as HP I'd go down the negotiating trail getting as much in writing as I could, before turning it all over the the authorities.
"Tru64 Unix for PA-RISC"?
Not surprised they dropped support for that - can't have been much demand for it at all :-)
Redwood Shores, CA, April 1, 2015
Oracle today announced it would be dropping support for all remaining competing hardware and software platforms, requiring all Oracle products run only under Solaris on SPARC. This most recent step was only possible due to Oracle's near monopoly position of 97% of the enterprise software business.
An Oracle spokesman was quoted as saying "Exadata 3 is all the server you'll ever need. Choice is not a very efficient business model for us."
Oracle and Europe
Rememeber when the EU made all kinds of noise about Oracle possibly behaving badly after they acquired Sun? They blocked the deal until Oracle promised not to mess around with MySQL. I wonder what the EU is thinking now about Oracle's actions regarding HP and Itanimum.
The EU could care less...
The EU probably doesn't give a feck, why would they when they are pretty much powerless to do anything?
The EU has plenty of power to investigate market abuse and raise fines if necessary. Even Microsoft were hit by this. Given that there is at least one very large European company (SAP) which has a lot to lose if Oracle abused their market power, then the EU could well take a lot of interest in this.
As anybody who runs a large IT shop can tell you, the costs of migrating from one database to another are enormous. The market is therefore quite "sticky" in terms of changing vendors.
re: EU Power
The EU is as impotent as the UN. They backed off of the Oracle/Sun merger, after huffing and puffing for a while, and Larry got everything he wanted with little or no concessions.
SAP and HP should hurry up and get on with it -- merge already! The US Justice Department will not have a problem, but I'm sure that the EU would be very upset to lose the largest European tech company to a US company -- or would it be a merger of equals like the HP/Compaq merger (LOL)?
Lets not forget
If it wasn't for Mark Hurd and his "inaccurate" expense reports
we would still be dicussing the same old issues.
I guess Jodie Fisher can add the Itanium scalp to her resume...
(not bad for a former porn actress and reality TV contestant)
Witness the driving force ...
"... to buy Red Hat for its operating system....".
Interesting "... to buy Red Hat for its operating system....".
But since HP does not realy have a end customer software stack, It would be more interesting to have a Microsoft and HP merger.
Then we would be back with the 'big' 3: IBM, Oracle, MSHP owning both a hardware and software stack.
Sticky DB users
The article states, rightly in my opinion, that companies are loathe to change their databases. Thus Oracle are in a winning position. However I would posit that acting like a complete arsehole towards your customers will mean your precious DB is unlikely to ever figure in any future projects, hence by increasing end-user support costs you are merely borrowing a small portion of tomorrow's revenue to tart-up your yacht today at the expense of that future revenue evaporating. Systems have finite lifespans and nothing focuses the mind quite like getting shafted - hence the Paris icon.
Seriously, outside of some telcos and some pharmas who used Itanium by the truck load? How many people actually used Itanium? Everyone I've talked to says they have some but not too many. Every hardware vendor except for HP has walked away from Itanium. Updates for HP UX have come out periodically, but I don't believe they've been as frequent as AIX, Solaris, definitely not popular Linux distributions like RHEL and SuSE, or for that matter Windows. When Microsoft killed support for Itanium a while back, there were minimal complaints. When Red Hat killed support for Itanium (but still has support for ZLinux on the Mainframe), but minimal complaints about that. When Oracle killed support for the Mainframe a little while back, where were the complaints about that? Why now? I believe it's because it's easy to hate Oracle because they're really good at what they do. But, surprise...they've done it for years. Why the sudden outcry now? Because they're too large? Why no complaints about IBM or SAP? Or are they just poor victims in all of this?
Microsoft isn't much better. They own all the corporate desktop environments and are expanding from their SMB space to becoming very competitive in HPC/Grid and large enterprises (for enterprise deployments i.e. where your IBM and Oracle type companies ususally play).
Is Google any better? They're building a data management monopoly from all the weather maps on almost every cable network I've seen in the US to owning search and advertisement to their revamped attempt at social networking (Google Buzz) to their Android based Java fragmenting technology, which is closed and bit off Sun Micro code. I'm pretty sure they'll own data on just about everyone in the world in a matter of years if they haven't already done so.
Everyone used to complain about Sun's technologies. Everyone complained that Sun was old, proprietary, etc. Forget SPARC, what about all their open source software technology? (For the record, SPARC wasn't proprietary when Sun owned SPARC. The SPARC instruction set is published as IEEE Standard 1754-1994. They also open sourced their chips on opensparc.net.
How many people actually embraced them and cut an check because they truly believed in open source? Does anyone remember Open Solaris, Open DS, OpenSSO, Java SE being GPL'd, Sun Grid Engine? How many customers actually deployed them and actually paid for its support? How many companies actually paid for MySQL Support despite all the increased investment that Sun put into MySQL to improve its scalability and create a non-Oracle owned IP based storage engine? How many people bought into the Sun Greenplum datawarehouse, which at the time was based on PostgreSQL, Solaris, and ZFS, esp as an Oracle/IBM alternative to keep costs down i.e. multi-vendor strategy? People forget that vendors like Oracle tried to make SPARC irrelevant despite very strong relationships in the past by relegating them to tier two or three or reduced or non-support. Sun fought like crazy to keep itself afloat by creating non-proprietary alternatives. In fact, I remember Scott McNealy came to town for an customer dinner and his whole pitch was about lower the barrier to entry and exit for customers and how open sourcing the technology, and complete indemnification from Sun to its customers would protect us as customers from acquisitions or a potential Sun bankruptcy.
How many companies and individuals posting jumped on Sun's strategy despite their rhetoric about "how %choose your proprietary vendor" is the devil?" Hell they even open-sourced their chip and had a few derivative chips created by third parties as a result. Obviously it was very few, because the Sun business model of open-source software with free licensing and paid support subsidized by hardware sales failed miserably because few people bought into it to sustain them any further.
What it comes down to, is your app and DB vendors (regardless of whether it is IBM, Oracle, ISV or boutique app ISVs) own your IT environment because you're scared of risking the stuff that makes you money in favor of some philosophical position. Sure world peace sounds great, but not if my taxes are going up or if my jobs are being outsourced, right?
You get what you pay for and large enterprise companies (who drive a ton or revenue that actually gains some profit as opposed to HPC outfits that drive a lot of revenue but very little profit) as a whole have opted to pay for traditional enterprise IT companies that can provide them infrastructure, some key business-tied application ISVs, a lot of in-house code, and a guaranteed support model. I think everyone hammers on SPARC (and given its more recent history, don't say I blame them), but the reality is that servers (regardless of vendor or architecture) are relatively minimal in cost of the entire IT stack. Software (Business, Virt, Management), Storage costs, facilities, and labor related costs are much higher than your servers. Servers just happen to be the most obvious and transparent way we spend money on IT. The one notable exception is Red Hat.
But Red Hat is a smaller company and there is a lot of IT vendor consolidation going on. Someone above stated that the key reason why everyone loves them is because of their hardware neutral position. I completely agree with them. I think their support is hit or miss. Sometimes you get an awesome person, sometimes you don't. They're not always adopting the latest and greatest kernels, but hey - they're commercial, I can get some sort of support to at least offer an internal support resolution SLA (or client facing), they're popular, they're growing and I'm not the only one doing it (i.e. I won't be fired for making the switch if something fails).
But with the IT consolidation going on, and the larger IT vendors trying to improve their standing vs IBM and Oracle, who are clearly the ones to beat from a complete stack perspective, HP, Dell, SAP, VMWare, Cisco, and EMC have all been either acquiring companies or creating partnerships. How long does Red Hat have? I don't have a crystal ball, but I can totally see the following companies trying to increase their respective value proposition to customers:
-Dell buying Red Hat because, they're just a low-cost hardware vendor. Even if you don't want them, people throw them into RFPs just to keep the price down.
-HP buying Red Hat for and OS/virt capability in the x86 space that can possibly expand into Itanium, WebOS spaces.
-SAP buying Red Hat to round out their OLTP database, BIDW, CEP, mobile computing, business apps, and virtualization spaces.
-IBM could possibly buy Red Hat for the same reasons as HP i.e. have a better value prop in the x86 space and have a one low OS option for all their tiers i.e. x86, Power, Mainframe.
Not saying it will happen, but there is room for it.
- Comment Renewable energy 'simply WON'T WORK': Top Google engineers
- Useless 'computer engineer' Barbie FIRED in three-way fsck row
- Game Theory Dragon Age Inquisition: Our chief weapons are...
- 'How a censorious and moralistic blogger ruined my evening'
- Leaked screenshots show next Windows kernel to be a perfect 10