"Crazy Larry, his claims are IN-SA-A-A-A-A-ANE!"

Magically Oracle RAC now scales to 8 nodes per rack * 8 racks = 64 nodes for OLTP?

We settled on 3 nodes for the applications we have not ripped RAC out.

"5 Terabytes Flash Cache Memory - Not Flash Disk!!!" Marketing BS. Too bad you forgot to change all the other charts in the deck that said Flash disk....come on you can do better than that...fire someone in marketing....along with that guy who can't forward the charts fast enough.

We looked at Exadata but even though "its the most successful product launch ever" it looks like only 9 companies bought/took one last quarter

* Apple

* Research in Motion

* Amtrak

* A large East Coast insurance company

* Thomson Reuters

* NTT

* Barclays Capital (UK)

* Two banks in Europe

I looked out at TPC-H and I did not see a new result for 1TB. Who does 1TB any more? I guess you want to play with the little vendors. And why did Oracle show the database vendor name and not Fujitsu and Sun? Oh and you did not show the HP blade result which your own team could not beat? That has got to Hurd.

8 cores per node * .5, 8 nodes per rack, 8 racks = 256 Oracle licenses of Oracle EE and RAC =

$16.5 Million for just the base oracle database for year one!!!!!!

"Crazy Larry, his prices are IN-SA-A-A-A-A-ANE!"

So no SPARC in this product, fair enough. But a Sun-branded solution that doesn't lead with Solaris? That's crazy talk.

Oh, and Fowler is an *Executive* Vice President at Sun. That sort of distinction seems to matter greatly to those in the thin air of the executive offices.

First you complain that Oracle is not proving their love for SPARC, so Oracle professes that they'll double down on SPARC... OK... So you bitch that they obviously don't care about x86 since they didn't say anything about it... So, Oracle and Sun show how much they also love x86... So then you bitch that they didn't also do something with SPARC!!!!

HEY TPM!!! IBM wants their marketing collateral back!

This latest Oracle/Sun announcement on Sun Exadata Version 2 now confirms that Oracle is also committed to Sun's x86 servers and storage!

Your comment that SPARC is MIA is clearly being spewed by your employer, IBM. Oracle has already confirmed several times that they will be investing heavily in SPARC.

Are the following not enough to confirm this?

Sun SPARC + Oracle is faster than IBM on TPC-C

http://www.oracle.com/features/sunoraclefaster.html

"Where in it to Win", "Spend more money developing SPARC than Sun does now"

http://www.oracle.com/features/suncustomers.html

"Are you going to discontinue the SPARC chip?" No. Once we own Sun we're going to increase the investment in SPARC. We think designing our own chips is very, very important"-Larry Ellison

http://www.oracle.com/sun/lje-oracle-sun-faq.pdf

http://www.reuters.com/article/marketsNews/idINN0740285120090507?rpc=44

"Oracle plans to grow the Sun hardware business after the closing, protecting Sun customers' investments and ensuring the long-term viability of Sun products"

http://www.sec.gov/Archives/edgar/data/1341439/000119312509103352/dex991.htm

http://www.sec.gov/Archives/edgar/data/709519/000119312509085803/ddefa14a.htm

@pierce they are 3.5" SATA drives.

@AC From what I could gather the prices included software. I can dig the term "Flash Cache" as the flash is mounted directly on the PCIe network interface card, not in drive bays. From the presentation I gather they're not addressed as disks at all.

As far as the system not being SPARC/Solaris, I think the few months that Oracle has owned Sun is too short a period for them to port the system to an entirely different ISA/OS. Probably next round though.

"If Oracle is trying to convince Sun customers that it is committed to the Sparc platform, perhaps it is not trying hard enough."

Or perhaps El Reg is just shitty for getting it wrong...

Ofcourse Oracle+Sun went X64. Its the place for price+performance. Lets face it: SPARC is dying/dead. It sure was cute tho...

Nice box. No SPARC, No Solaris.

So, Sun prixed for their chassis design then...

Firstly, I agree that this has nothing to do with SPARC or Slowaris, this is simply Oracle taking a developed version of the old Oracle-hp x64 design and slapping it onto Galaxy kit. At this point it's probably a bit early to expect Oracle to have a Slowaris on Niagara version, though the likelyhood is qustionable given that it would probably be rather price uncompetitive with their own x64 version. Nice to see that even the Sunshiners are having to admit there are cases where Slowaris-Niagara is just not the right choice and "commodity x64" is.

"......"Right up until the minute that HP starts partnering with IBM and Microsoft to do similar optimizations to take advantage of flash with their databases."

"Funny. Let's see how/if that plays out. My take: it won't...."

Really? You seem to have forgotten that hp have made a big success out of being the the equivalent of the industries biggest slut, being happy to jump into bed with anybody! Take a look at the hp websites, you'll find "strategic relationships" with everyone - Oracle, Microsoft, SAP, IBM Software, Red Hat, Uncle Tom Cobbley (OK, maybe not the last one!). There have only been a few times I can think of when internal politics kept hp from partnering up with someone, one case being support for Slowaris x86 on ProLiant. That was only stalled during Carly Fiorina's rule by the old hp guard, soon as Hurd got in it was "let's make some money selling those Slowaris guys some decent servers and support!" If IBM decide to make a database machine equivalent you can be pretty sure it will turn up on hp kit seeing as more IBM software is sold for hp kit than IBM's own (sorry IBMers, that came direct from a senior IBM salesgrunt, so I have no reason to doubt it). As for M$, since when have they turned down an opportunity to sell SQL Server licences no matter whose kit it was on? AFAIK, there is nothing unique or "locked-in" about the Oracle-hp database machine work that could stop hp offering the same kit for M$ to put their own version on top of, should M$ want to try it. If such storage appliances take off then I'd say we can positively expect IBM and M$ to be jumping on the bandwagon.

"First off, your addressing of Solaris as Slowlaris clearly shows where you come from and who do you align yourself with...." Strange as it may seem to Sunshiners, who seem to believe there is a factory somewhere that churns out "haters" ready-made, many people get their views from experience. My experience of Slowaris is that it has been and still is slow, even on the current Sun servers, compared to AIX, hp-ux or RHEL, which are the competition I have seen in action.

"....Solaris stopped being slow a long time ago, you should take it for a ride sometime....." I have taken Slowaris for many a "ride" and I still do see Slowaris at work regularly as we do competetive shootouts for all purchases. Unfortunately for you, Sun hasn't won any of our shootouts for a long while despite this having historicly been a Sun shop many years ago. This has nothing to do with whom I "align" with, it's a simple, empirical fact. Now, my experience can't be the same as everyone's in the market, but seeing as we're a large corporation running mainly mainstream apps (including RAC), it would seem pretty representative for the targets Larry wants to get into and Ponytail failed to.

As for hp needing a "competitive solution", you are forgetting who did the hardware bit for Oracle in the first version. If the market takes off then expect hp to want in, even if it is just offering the same Oracle stack on hp hardware (like those market-dominating blades for those Infiniband switches you mentioned). As it is, hp has the advantage of the massive economies of scale of being the lead x64 vendor whilst Galaxy is in the also-ran bracket. And hp doesn't really care who makes the software if hp gets the hardware deal, just look at the historical support for DB2 on hp-ux and MS SQL on Windows whilst at the same time hp was rabbitting on about their Oracle partnership. Since hp is not a database vendor, why should they care whose database gets sold, as long as the solution works to hp's standards and makes them money.

And best of all, should Red Hat or Novell or another Linux distie make their own competitive stack (maybe with a branch of the MySQL code - now wouldn't that be ironic!), hp will be primed to sell that, unlike Larry and his on-off Linux relationship. And before you Sunshiners start saying no Linux distie could ever do that, I seem to recall many of you sprouting the same years ago when Linux was first mooted as an enterprise server platform. Simply saying "no-one else can do this" is the same head-in-the-sand posture that saw Sun go from the market-leading UNIX vendor to market-trailing cripple.

/SP&L

@Scott K. Handy; When a VP from IBM make comments on ElReg he must be really scared of what's coming from the competitors.

A must read for all Oracle and Sun customers

http://www.businessweek.com/magazine/content/09_38/b4147052120632.htm

Sun was -11% in Q2, -20% in 3Q, -30.6% in Q4 (minus 31% in their year end quarter?)

Mr IBM Veep, let me speak a word of fairness being both a IBM Power and Sun SPARC customer.

I've had the privilege of hearing of what Sun has planned and delivering for flash.

Mr Veep, Flash is not just SSD. IBM is only thinking of SSDs in Power or Storage.

Sun has got a whole eco-system going and it is nothing but spectacular. Breakthrough economics is what I see. Plus how ZFS is using flash technology (not just SSDs) is simply amazing.

I had my IBM rep talk about how shitty Sun is doing, how the Sun is setting, quite similar to your tone and language. Two words Mr Veep -- "low class". My feedback, I appreciate honesty, I appreciate good service, pro-activeness, I love great technology, I love saving money for my bosses. I hate mud-slinging. Just too unprofessional, not what I would expect from a corporation like IBM.

I'm staying put on my Power systems and my SPARC systems, but all these moves that Sun is making with Flash is making SPARC exciting again, and I'm very open to adopting this, especially if Oracle is having a solid result on these kit.

BTW, your rep must be pretty desperate for numbers for him to predict the sun crashing and I will have no sales and support. The vibes I'm getting from Oracle on the hardware business are good. So I'll be patient and wait for the October the 14th on how Sun SPARC system can beat the IBM Power 595.

So are all the other Unix vendors... Also, the EU is taking their sweet time to approve this move, so that's making Sun's customers hold off on purchases. Keep your FUD to yourself. We already get enough from TPM and MB. Let's work with facts here!

When Oracle bought BEA they told me my Oracle application licenses would not transfer to WebLogic and the Oracle application suite was being sunsetted for WebLogic so I better start moving.

The next week Oracle raised WebLogic prices 47%.

The only way I got my 60% discount and 18% maintenance of purchase price (not list mind you) was to start bringing in IBM technology.

DB2 for my BI applications and Power to reduce my license counts and virtualize to get value out of my licenses.

I am sure most of the Sun gear with be "Sunsetted".... Larry isn't crazy he is buying a captive install base...well I am not your captive since I have IBM fighting for my business. With the new Oracle extensions in DB2 we are finally looking at possibly dumping Oracle all together. For now I have just focused on reducing my license counts with hardware from the blue company to replace the purple companies gear to save on the red companies extortion like pricing

"...factory somewhere that churns out "haters" ready-made,"

Nope, just you Matt. No one complains about other "haters". You are the only one that spread FUD to the extent that you do.

Timothy Prickett Morgan writes, "If Oracle is trying to convince Sun customers that it is committed to the Sparc platform, perhaps it is not trying hard enough."

That is why there was a full page October announcement about SPARC scheduled before this Intel announcement!

I am not sure what else was wanted!

There was plenty of speculation that the Intel boxes would be sold off once Oracle acquired Sun.

This September product announcement puts that to rest.

> My experience of Slowaris is that it has been and still is slow, even on the current Sun servers, compared to AIX, hp-ux or RHEL, which are the competition I have seen in action.

Funny that. I'll bet DOS 3.2 is even faster than RHEL on that hardware. Doesn't do as much, though. Just like RHEL compared with Solaris.

Sure, some people don't care about functionality or reliability, and just want benchmarks. The professionals know otherwise.

@Scott K. Handy;

http://www.channelregister.co.uk/2009/09/16/idc_emea_servers_q2_2009/

>>> "So I'll be patient and wait for the October the 14th on how Sun SPARC system can beat the IBM Power 595."

> Did you mean Sun SPARC system or 10 Sun SPARC systems?

> It's obvious that 10 fastest T2+ servers in RAC cluster will be faster than single 2-year old IBM machine. IBM or HP could do the same and connect 10 superdomes or 10 595s in RAC cluster to achieve higher scores. All your money would go to Oracle then...

Come'on why all this mud-slinging? IBM does 32-node clusters with 2 chip P570 for TPC-H too, right?

I'm not exactly clueless about IBM and Sun performance, especially I have large scalable applications on both platforms. So stop this bashing, and I only trust the performance data I'm monitoring on my boxes.

For the Power 595, come'on, every technie knows that though IBM used a 64 core Power 6 with 5GHz processors, you had like 11,000 spindles with crazy database partitioning. You always claim 2-3x per core performance over SPARC64, but in my experience, it is about 1.5x per core at best. Qualification: Common storage on EMC.

T2+ processors have been incredible with all my throughput applications. And surprisingly IBM Power could not keep up with the throughput and latency was bad at high connections.

So mudslinging aside, I love technologies from Sun and IBM. They are different, but that brings value to me. I see IBM as a steady improver, but Sun dares to think out of box.

RE: Re: RE: @Matt

"....Nope, just you Matt. No one complains about other "haters". You are the only one that spread FUD to the extent that you do....." Lol, so anything anyone says that you don't agree with is FUD? Nice, narrow set of friends you have there. Next I'm sure you'll be claiming we're the only customer that has ever migrated off SPARC Slowaris. All the more funny considering my first post was actually a rebuttal of the idea that the new Exedata being Galaxy-based was somehow a disaster for SPARC. Or are you saying that was a lie too?

RE: Mattie Pattie Laddie

"So Solaris is slow?...." Yes. In straight shootouts, same generic app stack, same rough price-range, Slowaris 10 vs AIX or hp-ux or RHEL, Slowaris was usaully the laggard. In all cases Sun and their resellers were allowed to tune the kit as they saw fit, so there was no chance of those builds not being the best available. The only instance where Slowaris outperformed the others was on T2 on webserving, but then it lost out on price to RHEL on ProLiant. You can squeal all you like, but that is my direct experience, and you really don't have either the experience or capability to convince me otherwise, especially as I susepct you have never even seen half the kit discussed here.

"....If I consider your (proven) earlier lies...." The only thing you have proven is that you have neither the experience, knowledge or debating skills necessary to join these discussions. At this point, all you generate is a mixture of unintentional humour and laughable immodesty. Please do explain to me how a maths degree makes you at all qualified to discuss computing, when probably more than half the readers here have computing based degrees and a shedload more practical experience?

RE: Slowaris?

"....Sure, some people don't care about functionality or reliability, and just want benchmarks. The professionals know otherwise." Nice attempt at a feature sell without any substance to back up your insinuation that Slowaris has "unique" faetures that Slowaris can provide. Just to forestall the inevitable Sunshiner post of all the "unique" features they calim make Slowaris soooooo much better than anyone else's OS, let me explain that we have, as part of our accpetance criteria, a long list of features and capabilities that solutions must meet before they go through to the shootout stage. As it stands, RHEL, AIX, SPARC Slowaris and hp-ux provide all the features we use, some better than others, but all to the standard we require. Where Slowaris fails is on price (even after discounts), reliability (my experience), and performance. Of course, there is also the board's lack of belief in Sun (a truly unique feature).

But, just to cheer you all up, Exedata is on our list of systems we will be looking at in 2010. So there is a chance we may actually buy some Sun servers. Of course, if Oracle release a version on our preferred ProLiant blades then the Galaxy option will likely fail without some serious discounting.

/SP&L

"Lol, so anything anyone says that you don't agree with is FUD? Nice, narrow set of friends you have there. "

No it is merely because you claim to attack Sun, like you do, because of all the FUD that everyone spreads about HP. You rarely provide anything concrete for your arguments. You merely state that Oracle will in no way keep Sun's hardware business and never intended to keep it. This is the definition of FUD. You never lose a chance to attack Sun or anyone that defends or applauds Sun's actions. Yes, you are a FUDster and no, when someone says something specific about Sun that I disagree with it does not automatically make them a FUDster. You have proven your intentions and as a matter of fact you have said that you do not like Sun. You will say anything to convince others and you spend an inordinate amount of time trying to convince others that Sun is evil and anyone that likes them is an idiot. Go back to your mothers basement where you belong.

...and you wonder why everyone thinks you work for HP or a HP reseller.

Fact: Memory management in Linux sucks

Implementation is reminiscent of solaris 8. (see linux hugepages, and try to find something like ppgsz on Linux)

False statement: Linux is cheaper.

Solaris ZFS compression allows you tu build faster solutions with half the number of disk drives. reducing initial and operating expenses.

Linux's lack of binarry compatibility will cost you dearly, adding a lot of aditional work and complexity to your upgrades. Saving a few pennies will cost you dearly.

We've been using both Solaris and Linux on our servers. And I have to admit, I am not impressed by Solaris. Guess what servers crapped out when power went down (the filesystem was all corrupt). I have to admin it was using UFS, but again guess on which server I saw Oracle reporting corrupt blocks (on an x86 with Solaris and ZFS). Upgrades are headache, lack of official software repositories, installation problems even on Sun Servers. We are using Oracle a lot and deploying on ASM does the job perfectly. When you want to install an application (that is not int the top 20 most pupular applications) guess on which OS you have to bang your head to compile it !? I will try to stay away from Solaris, except headaches we never got something good. And with the binary compatibility your areimpacted if you write really low level code (drivers and stuff), so zvonr dont spread you FUD around.

@ares

sorry, but your comments sound a lot more like lack of skills from the administrator rather than any disadvantage from using solaris.

There are repositories for Solaris and OpenSolaris, like Blastwave. Granted, they are not official, but Solaris 10 gets updated every six months or something. Remember it's an stable platform, not bleeding edge like Fedora or Ubuntu.

Binary compatibility has nothing to do with low level code. Nothing at all. It's about an stable ABI and API that allows code compiled (binaries) from the earliest versions of Solaris to run unmodified on Solaris 10.

And the fact that a lot of OSS software is Linux centered doesn't help when trying to compile some random program (read, not used by a lot of people) on Solaris. I assure you, it is easier to do so on OpenSolaris, give it a try.

@TPM

It's funny that your review of Exadata v2 tries to make it look as something unimportant, yet on your itjungle article here: http://www.itjungle.com/tfh/tfh092109-story02.html . you say that IBM should do it's best effort to copy it.

So, it's a good idea if IBM came up with it first, but not if Oracle did it?... Then which is it, is it good or not?

Oh, TPM, you IBM troll... make up your mind, please.

You say IBM "could" do this 13 years ago... As I said earlier, if it was SO obvious, why didn't they do it?... Haven't you though that they might not have found it viable?... Too costly/ too slow to sell it to anyone?

coments like:

"Each storage server is equipped with four flash disk modules, each weighing in at 96 GB and apparently plugging into PCI-Express peripheral slots. (Sun and Oracle tried to make a big deal about this, calling it "FlashFire technology." But i 6.1 is already flash aware, as you know, and can move data in the single-level store for memory and disk into flash of that will boost performance.)"

Are just sad... why does it hurt you so much that Sun + Oracle did something good and you try to show it as "no big deal".

Oh, and managing Oracle RAC clusters isn't nearly as hard to administer as you imply.