Oracle defies the economy – and the curse of Sun

Declining hardware sales, a rocky global economy, and a tough compare with the prior quarter did not slow down Oracle much in its first quarter of fiscal 2012 ended August 31. But these factors did slow it down a little. However, it was not enough to keep Oracle's profits from rising nicely as the company wrings costs out of the …

[Allison Park]

100 customers per quarter

Considering how many loyal Sun customers there were 100 new exadata OR exalogic customers per quarter two years after announce is pretty lame.

I have seen Oracle be pretty cagey about the numbers but how much of their hardware revenue is exa-crap?

Everyone I talk to wont get near an m-class since it is 5 year old technology with no future SPARC64 technology.

As far as the T4 which seems to be the only future for SPARC I saw Oracle just doubled the price even though customers will be lucky to get 2x the performance per core. Just before we dumped Sun technology off the standards list we tested the virtualization on T and found the threads to be tied to closely to the cores which made it no better than the ldom technology of the prior generations.

Hat's off to Oracle for driving our Sun costs and reaping software dollars, but I still see a rapid retreat of Sun customers from Oracle hardware and especially the lock in of exa-money systems.

Cheers, Allison

[A Non e-mouse]

Hardware Margins

"gross margins on hardware rose to 54 per cent, up from 48 per cent"

50-odd percent hardware margins ? That sounds high. I thought only Apple could manage hardware margins close to that ?

Which also begs the question, if margins are so high, how come Sun got into such dire straits ?

(Genuinely asked in ignorance, I'm not a financial whizz)

[Gordon 10]

That's the billion dollar question.

The second one would be how have oracle managed to drive the ex-sun costs down far better than sun ever did. Does it just come down to being more ruthless?

[Tomato42]

It's easy if do drive your costs down if you only leave box movers and marketdroids to sell the boxes.

It's not a sound strategy long term, but the exec will long leave the company before that becomes visible...

[John Riddoch]

To be fair, you need high margins on low-volume sales to make it worthwhile. If you're selling 5 million 2 socket x86 boxes, you'll drive an overall profit on a 5-10% margin, but on a high-end box where you sell 5000 unit, you need to divide all your R&D across fewer boxes and consequently a higher margin.

Add in that very few people buy at list price (40-60% discounts are common on high-end servers) and it's not quite as rosy as it might seem.

[A Non e-mouse]

"very few people buy at list price (40-60% discounts are common on high-end servers)"

But surely the reported 50% margin is on what was sold, not on the list price ? Which makes the list price of SunAcle servers out of this world.

[Jesper Frimann]

RE:Hardware Margins

Woow 54% gross profit margin on Hardware ?

Now that is alot, that is what 14% more than IBM, which has a much larger portion of it's sales in the highend than Oracle has, where the profit margins are higher.

They surely must have cut to the bone on sales, and/or charge to much for their kit.

// Jesper

[Anonymous Coward]

how come Sun got into such dire straits ?

Management by ponytail.

[Ken Hagan]

Not caring about *his* x64

There's a difference between "I don't care about the x64 systems that *I* sell because the margin is tiny" and "I don't care about the x64 systems that everyone else sells (with tiny margins) and which represent 99% of the market for my software (which has huge margins)".

The difference matters because the latter position doesn't imply that he cares about his non-x64 hardware. Yes, he's got more of his own IP in them, but he only *has* that IP because the company that paid for its development went bust. If he has any sense, he'll bear that in mind.

Until someone comes up with a convincing variation on the von Neumann architecture, hardware is a natural monopoly because alternative ISAs just aren't different enough for their performance benefits to outweigh the cost of their own development.

[Anonymous Coward]

Pray tell

Please explain the rise of ARM then? The company behind it does nothing but figure ways to make it better and doesn't even make hardware themselves.

Besides, the selling point (as detailed here earlier) is that a full sunacle stack is (or well so say their calculations, of course) overall cheaper than an oracle-on-someone-else's-x86 stack, even with fatter hardware margin. And for that to work well as in being believable, it doesn't help to have to buy the cpu from a low margin mass producer like, oh, intel or amd.

And then, sparc is a reasonably well thought out ISA and indeed does have a couple nice plus points that have lots of speedy potential, but the thing could use a bit of spit and polish. Sun ran itself into the ground and didn't have the money left to stay competetive. Oracle doesn't have those cash problems. Handled well (ie quite entirely unlike java) it could turn into a community binder, too. We'll have to see what they do next but depending on how you look at it their hand is not all bad. If they make it work for them, not at all.

[fch]

SPARC features ...

... were reasonably well thought-out when the architecture was designed; but since you're comparing it with ARM (not a good comparison,what do electric bikes and humvees have in common ?), I'd like to comment a bit on that.

ARM have continuously evolved / improved their core instruction set (ARMv5 -> v6 -> v7, all adding quite generically-useful things), and that even though few people would call ARM's initial instruction set design anything else but "great". ARM and its licensees also take hardware advances (caches, builtin ram, close-coupling between CPU and devices) and continuously incorporated such into the implementations. Anyone can _see_ how ARM leads when it comes to CPU instruction set design / improvement, or, to phrase it differently, achieving the "speedy potential".

SPARC is the monolith of instruction sets, though - set in dark, menacing, impressive stone since sparcv9 was concocted (1994 ?). Some of the things SPARC does (branch delay slots, /dev/zero register, ASIs, the instruction set extensibility) are indeed useful. Others - fixed-size register windows, or even the windowing mechanism as a whole have proven to be more of a burden; they make SPARC programs use larger stacks and therefore require comparatively larger caches to achieve full potential. Then there's Sun insistence over decades not to consider out-of-order implementations, even at a time when Fujitsu's SPARC64 had already proven the usefulness (fortunately, the T4 finally addresses that). Also, again in comparison with ARM (or x86), SPARC machine code isn't very dense - larger code footprint, again needs bigger cache / cache bandwidth. There would be ways to address that (like x86's micro-ops caches), so agree with your assessment "speedy potential", yet a lot of that remains unexploited.

Also, very unlike ARM where there are a variety of widely-used instruction set extensions that ARM keeps on improving regularly, SPARC has not had any updates on that front for a while either. Yes, the T-series have the crypto accelerator as closely-coupled device, but that's as far as it went. Not every design considered "great" will keep that label over time; at least from my point of view ARM has done better there than SPARC.

SPARC is great because it's reliable, got very predictable behaviour, runs all your old stuff.

But SPARC has had great potential in the 80's, had great potential in the 90's, has been having great potential in the naughties and still has great potential ... and it will always have great potential.

Fingers crossed for the T4; some achievement instead of potential would be wonderful.

[Zobbo]

Oracle have bumped those margins up - IIRC Sun's markup was 40% in the early noughties.

[Kebabbert]

Oracle betting on clusters

Here is a world record where a cluster of x86 servers beat the biggest IBM P795, with a SAP world record result:

http://blogs.oracle.com/BestPerf/entry/20110915_x4800_m2_sap_sd

Larry has also explained that clusters is the future, not big SMP servers.

[Anonymous Coward]

Parallell vs normal SAP benchmark

As you should know if you have any experience at all, you should never ever compare parallell results to normal results.

The benchmark from oracle was 6 servers, each equipped with 80 cores, which gives 480 cores. The IBM solution was a single 256 core server, still the sun solution was only marginally better.

Sure the cost of 6 intel servers is lower in TCA, but in overall cost for the solution when it comes to sys managment, uptime etc. a unix solution built on more reliable hardware usually is cheaper.

If you want lower cost with a AIX solution, select smaller servers and scale out - the reliability will be better than x86 anyway. Of course the cost for the really large big iron from IBM is quite high when it comes to just buying one - but you'd be surprised of how well-priced servers such as Power 740 is. I would recommend a cluster of 740 servers to a large 795 any day of the week.

[Allison Park]

Parrell vs. single SMP

Kebabbert....don't insult our intelligence

Of course Larry's future is clusters not simple scale up. He makes about 3X the money adding on RAC and all the performance monitoring tools to make it work.

Comparing a cluster of 6 sun (cough) intel servers with the reliability of about 5 years with a mission critical enterprise server is not a fair compare.

Obviously Oracle needed to do a benchmark without a price comparison cause those 480 cores with all that Oracle clustering software and management tools would be insane.

Yet another example of Oracle finding a benchmark where there have only been 1 result in the last 3 years and trying to claim leadership, then comparing to leadership

Looks like they didn't want to compare the 4 node cluster with 40,000 users (about 10,000 users on each node) to the single node IBM Power7 at 15,600 users (almost 2 year old benchmark).

Oracle chose a very old (2008 non-unicode) prior generation benchmark to compare.

First rules of benchmarketing....the marketing department is in charge.

[Kebabbert]

@Allison Park

"...Kebabbert....don't insult our intelligence..."

I dont get it. Have I tried to insult you in one way or another? Have I tried to make it look like you are an idiot? No, I have only posted some benchmarks. What is your problem with that?

Sure, it is a clustered benchmark vs a IBM P795 - but I were careful with that and pointed it out. I did not try to make it look like it was benchmarks on equal footing. I just posted a benchmark, why do you feel insulted by the fact that a cluster of x86 wins over the biggest IBM P795 in SAP? I dont understand? Everytime IBM looses a benchmark you feel insulted? Are you serious?

[Jesper Frimann]

Re:Oracle betting on clusters

Sure they are... If you read the benchmark text then the SAP app code ran on the

Now what does an Oracle Enterprise edition with RAC cost running on 480 Cores ?

Lets see the per core factor for the cores is 0,5 with 480 cores.

the cost of the Oracle license is 47,500 $

the cost of RAC is 23.000 $

Hence the cost is 16.920.000 $ + 3.722.400$ per year in support. And that is just for the database.

And we haven't even started piling up other database features, like DB management etc. and hardware and ....

This is why Oracle is pushing clusters.

Furthermore SD2 parallel is well, almost an irrelevant benchmark. Why, cause it's a embarrassingly parallel benchmark run on a cluster.

Each node will be able to pull around ~25.000 users at 99% utilization. if we look at the other vendors's 8 socket boxes. Now the Oracle submission runs at 92% utilization. Hence ~25.000 x 6 x0, 92 =138.000 is the theoretical max performance for 6 nodes. Now the Oracle benchmark delivers 134.080 which is 97%+ of the theoretical peak performance at 92% utilization. Geee..

Now if they did a SAP SD3Tier parallel it would tell you something about the DB layer. Not the 2Tier, that is just marketing bull.

@AC with regards to the POWER 740 being the best POWER b ox, nope use the 770. If you have a POWER 570 based POWER6 then a MESS upgrade to the 48 core version is actually a good deal, but remember to get a good discount. Don't let IBM charge you the same as for a 64 core upgrade.

A 48 core machine will still run out of RAM, if used right, before it runs out of processor power IMHO.

// Jesper

[Kebabbert]

@Jesper Frimann

It could happen that Oracle's solution is more expensive, I have not checked that. I trust you on this, without requiring any links from you.

My main point is that in a discussion of x86 vs POWER/SPARC, I wanted to show that x86 is catching up fast. In some cases x86 is actually fastest in the world today. A few generations ago that would not have been possible. This trend is VERY relevant when we talk about x86.

I understand that any benchmark that shows superiority over POWER will make lot of IBM supporters here upset (for instance Allison Park). But sorry, that is how it is. x86 is coming, and nothing can change that. Ivy Bridge version of Westmere-EX will be 40% faster, and hence, x86 will be faster than POWER7. Sorry for posting that. Dont get too upset.

But, one thing that x86 does not have, is RAS. x86 is buggy and bloated and sucks really bad. If all that investment in R&D had gone into POWER or SPARC we would have much faster and more energy efficient cpus than todays x86. x86 will never rival POWER/SPARC in terms of RAS - it is too buggy and ancient and ugly.

[Jesper Frimann]

@Kebabbert

I must admit I fail to see what your clustered SAP SD 2-TIER benchmark has to do with x86 scalability.

But perhaps you should have a look at your good friend Larry's statement about his x86 business, before you get all hyped up on x86.

http://www.eweekeurope.co.uk/news/oracle-happy-to-see-its-low-end-business-tumble-40363

With regards to Ivy Bridge. Then there seems to be a big difference between what is released, and what you think is released. You won't see a Ivy Bridge -EX processor in 2012 if INTEL keeps the normal pace of releasing -EX models after -EP models.

Furthermore POWER7+ will be the processor shipping POWER servers when Ivy Bridge enters the marked. But hey, shouldn't you rather do your normal comparing against POWER6 HW from 2007.

// Jesper

[Anonymous Coward]

Kebabbert

I dont understand? Everytime there is a story about servers, you post about how bad System z is and quote Hercules MIPs numbers. Are you serious?

And no, I'm not Allison Park. Just someone else who laughs at your ridiculous posts.

[Kebabbert]

@A.C

"...I dont understand? Everytime there is a story about servers, you post about how bad System z is and quote Hercules MIPs numbers. Are you serious?..."

This is a story about servers, and I have not posted about how bad System Z is. So I dont understand your question? What are you talking about.

[Kiralexi]

Kebabbert

Anonymous Coward is right. You seem to like posting any kind of meaningless numbers that make Oracle look good. What are you trying to accomplish?

[Kebabbert]

@Kiralexi

"...Anonymous Coward is right...."

As I explained, he is not right.

.

.

I can explain again though, that I dont like all the FUD against Sun and now Oracle. I dont like lies, as you know. And then I counter the lies by posting links that prove the opposite. And when I do that, guess who gets all complaints? The FUDer or me? I dont understand how you can let all the FUD going on, without reacting? What is this called, hippocrasy or something?

I would not be surprised if you are the same FUDer that posts anonymously trying to discredit me. Why would I not be surprised? Well, I have NEVER seen anyone anonymous complain about all the FUD and lies going on here. Never. But all those anonymous persons are always complaining when I post benchmarks that show how superior Oracle is. That can not be a coincidence.

The absolutely dumbest attack on me, was when someone wrote something like:

"because of your posts here, Kebabbert, our CEO has stopped buying Oracle gear, and because of you, we have changed our IT-strategy and started a migration from SPARC to IBM POWER, that costs many millions. Please stop posting Oracle benchmarks. Everything is your fault. Stop it. Now because I really love SPARC but sadly POWER is better". I really smiled when I read that. It smells desperation, dont you think?

Regarding this Oracle benchmark I posted, I thought it was relevant because it shows how much performance x86 servers can deliver, winning POWER7 servers, including P795. This is a relevant piece of fact, in an discussion about x86 vs POWER/SPARC. We see that x86 can in some cases be faster already today.

In fact, POWER7 is only 10% faster than Intel Westmere-EX in some benchmarks. Soon the Ivy Bridge version will arrive, which will be 40% faster than Westmere-EX according to Intel. Then x86 should be faster than POWER7. And much cheaper. Is this not relevant for the discussion? No? So anything that shows how fast x86 ramps up performance should not be told, so nothing can threaten the POWER7?

[Jesper Frimann]

"Regarding this Oracle benchmark I posted, I thought it was relevant because it shows how much performance x86 servers can deliver, winning POWER7 servers, including P795. This is a relevant piece of fact, in an discussion about x86 vs POWER/SPARC. We see that x86 can in some cases be faster already today."

The problem Keb, is that given a benchmark that is enough embarrassingly parallel, a room full of mobile phones could beat a POWER 795. Hence the relevance of your comparison. There is no doubt that the Westmere-EX boxes and POWER7 based machines shipping today are really the two top dogs to look out for in the server business.

But they mostly do play in two different areas of the server marked, POWER in the midrange to highend, and Westmere based servers in the lowend to midrange.

Many years from now there will still be Itanium based servers, Mainframes, SPARC and POWER servers running. So perhaps you should cut down on the rhetoric, if you want people to listen to you.

// Jesper