How long will it be until Oracle stops selling systems that are based on Intel's Xeon processors? Or servers that use Fujitsu's Sparc64 line of processors, for that matter? To hear Oracle cofounder and CEO Larry Ellison tell it, tomorrow would be just fine to cease and desist making and selling x86-based machinery. The only …
He's coming for you...
This is going to strike horror into the heart of many an Oracle customer. They can surely feel the ropes tightening around them as expensive proprietary hardware becomes the norm. At least with x64/x86 support, there was some choice of suppliers and competition. If that ever goes, and Oracle becomes the vertically integrated company that Ellison appears to wish it to be, then that's where IBM was back in the 70s.
I see a backlash coming - from Oracle's customers.
In July Oracle posted an updated roadmap which had the next M-Series moving to 2013 the last M-series moving to 2014 and the combined SPARC moving to 2015. I wonder how they got them to all move back in. I think I read somewhere that Oracle's roadmap is now on their fiscal year vs. calendar which gives them 4 more months in a year.
Now in September they redid the roadmap to blur the years. What i have heard is the next M-series was supposed to be a scalable variant of the T4 but Oracle won't be able to get the T scaling beyond 4 sockets any time soon so they convinced Fujitsu to do another fab of SPARC64. I would say I was wrong that SPARC64 is at end of life...but that was their plan.
I haven't heard if the next chip will be on a new fab or still on the current 65nm. They sold their Fab to toshiba so not sure how that would work out.
It looks like they are still targeting the last M-series to be a T chip and the new one below is just the 4 socket version.
Looks like they are buying time till 2014 and hoping the continual decline in SPARC levels off.
It is good to see competition in the market.
Keep in mind that "In test" does not mean something is coming out soon. The ROCK chip was canceled in June 2009 but was in test in early 2007.. Apr 11, 2007 – Jonathan Schwartz has posted on his Sun blog that the Rock (the codename for Sun's upcoming processor) has arrived at his desk. Been there done that.
Well off to cancun.
Oracle RDBMS Pricing
Oracle should stop faffing around with SPARC and sort out their pricing.
Oracle is getting silly with its prices. Far more expensive than DB2 on the same platform. DB2 is actually cheaper than SQL-Server in some configs.
Even on SPARC hardware the support costs for Oracle is getting silly.n
db2 is cheaper only on ibm power where less than 4 sockets and on laid off systems.
oracle EE+parttitioning:(47500+11500)*0.25*2*16*1.22 = $575,840.00
db2: 405*16*2*70 = $907,200.00
4-sockets 10-core xeon EX
oracle EE+parttitioning: (47500+11500)*0.5*4*10*1.22 = $1,439,600.00
db2 EE: =405*10*4*100 = $1,620,000.00
The future is bright!
As I said earlier, Oracle's future is brighter than ever. My prediction seem to come true.
Regarding HP, I think that HP made a mistake by letting Intel learn how to do high RAS cpus. If Intel chooses to let Itanium go, then HP has no choice than to accept it. HP is now vulnerable, that is not good. Itanium is a good cpu and I wish it stays. I wish HP took control over Itanium and ramped up the pace.
To sell and control an entire stack from hardware to software gives huge benefits; performance and you can charge a premium. We see the result now. This benefits Solaris, and benefits Oracle 11g.
If Oracle chooses to stop develop the database on AIX, then it would be a serious blow as well. I dont know how many Oracle databases run on AIX, though. I heard that Solaris is the most common platform. So maybe it doesnt matter if Oracle stops development of the database on AIX and POWER.
Anyway, the roadmap seems interesting. I am not surprised that Oracle earns money. :o)
I have a feeling
that if Oracle stopped developing their database product, or in fact any of the products they have mopped up, on AIX, especially if they had already shut down Itanium development, there would be one hell of an anti-trust lawsuit in the US.
Plus the fact that they would lose many millions of dollars of revenue from those customers who like AIX and Power more than Oracle, and would switch to DB/2 rather than to Solaris/Sparc.
The problem here is that many applications are database agnostic, using ODBC and JDBC and SQL amongst other abstractions as the means of using a database, which allows them to switch database products relatively easily.
Many Oracle DB customers get regularly annoyed by them because they can't appear to decide what the licensing model is. Some customers I worked with ended up re-negotiating their Oracle license fees every year because the way it was worked out changed each year.
It seems that what I had been saying for years has finally been proven right: HP fucked up abandoning PA-RISC and betting on Intel's Itanium. But then I think that one can be chalked to Carly's Follies...
<Sigh> Some people just don't have a clue.
".....HP fucked up abandoning PA-RISC and betting on Intel's Itanium...." Itanium was hp's original work right from the start.
"....But then I think that one can be chalked to Carly's Follies..." The whole Itanium decision kicked off long before Carly's arrival at hp, and was a simple engineering decision - the hp lab boyos could see the finite limits of RISC in general and designed what they saw as the replacement.
I suggest you go do a little reading before your next display of ignorance.
Larry said, before he bough Sun, that Oracle DB is most common on Solaris. I dont know how common Oracle is on AIX, but clearly less.
"...The problem here is that many applications are database agnostic, using ODBC and JDBC and SQL amongst other abstractions as the means of using a database, which allows them to switch database products relatively easily..."
I have heard the opposite. Many applications are locked to a particular database, which makes it difficult to migrate to another vendor. This is very costly to do, and regularly I read about large customers that try to migrate to another database and fail because it costs too much. Thus, I dont think you are correct on this. So, if Oracle choose to stop Oracle DB on AIX too, it would be difficult for customers to migrate to another DB. The least disruption would be to buy T4 and continue on Solaris.
But I dont think Larry will kill Oracle DB on AIX. Why did kill Larry kill Oracle DB on HP? Well, because Itanium is going to be killed by Intel sometimes soon, Larry said. I have not heard IBM saying that POWER is going to be killed sometimes soon - because yet POWER is faster than x86. But I dont think you have to worry about AIX, because IBM has not yet given a date when AIX will be replaced by Linux.
Oracle on Solaris may be a majority,
I don't have the figures. Maybe you could post references. But I never said that it was mostly on Power/AIX, just that there was a lot of it on Power/AIX.
Everywhere I have worked in the last 15 years, with the exception of my current contract (who run Oracle on zOS and Linux on zSeries, not that I have any involvement in those systems) have run Oracle on AIX as their main DB. Quite a lot of it actually, and ended up paying Oracle big bucks (or, in-fact, pounds) in license and support fees.
This has been in (large) financial, government, utility and construction organisations.
I have not heard of many applications that were locked to Oracle unless they were Oracle apps (not a big surprise). Oracle may be the recommended or the best supported DB, but any 3rd party application developer would be limiting their market if they were unable to sell into non-Oracle sites. Oracle is not quite a monopoly.
I agree that it would be a costly and disruptive operation, but then so would replacing the servers, changing your management infrastructure and re-training your support staff. Bit of a no-win situation for anybody, so let's hope it does not happen.
Larry Ellison himself said that Oracle DB is most deployed on Solaris. He should know. He said this when he was buying Sun. He wanted Java and Solaris, he said. And added that Solaris is the most common platform for Oracle DB. I do not have any numbers, though. I just read the interview
Larry never said he wanted to get out of the x86 market, he said he wanted out of the low-end market where they have no value add. That means Exadata/Exalogic will still be around. I'm sure he would love to have SPARC replace x86 here, but SPARC will never have the volume to replace x86 cheaply enough.
As far as the next M-Series Proc, I would be willing to bet that it will not be a SPARC64 variant. Fujitsu has not mentioned anything about it, and Oracle has screamed about running everything on Oracle tech. Perhaps Fujitsu will be a reseller of Oracles M-Series tech moving forward? Any bets here?
I don't think that saving a small amount of money on the hardware on Exadata/Exalogic, by using x86 machines rater than SPARC based boxes is an issue.
T series servers most likely aren't that expensive to make, and it's all inhouse hence you won't be paying profit to Intel and others. Hence the price difference will be much much smaller than one might think.
I think one of the primary reasons why Exadata/Exalogic are still x86 based are historical and practical reasons. It takes time and effort to move to another platform.
Personally i think that if T4 or it's successor performs well enough, and their roadmap for SPARC is long enough, then it makes great sense for Oracle to migrate these solutions onto SPARC.
For example I think that IBM made a mistake when something like the SVC and SONAS was implemented on x86 rather than POWER.
What makes you think that the new M-series part is SPARC64? I don't recall them mentioning that in the announcement event, and it seems just as likely that it's a clock-boosted (4-5GHz, perhaps) 8-core 28nm T4.
The new M series processors are all developments of Fujitsu's SPARC64. The T series just can't compete - and never can - on single thread clock speed. Because of the way that T series despatching works it's virtually impossible for it to emulate al the single-thread speed-ups required for a super-scalar processor. Even clock-for-clock without core contention, the T series is only about 30% of the per-thread throughput. Sometimes it's a lot worse.
Unfortunately for the proponents of T series much application code out there is not amenable to parallel running.
re: Steven Jones
You must not have read the article. It clearly shows that the T4 is faster than the SPARC64 now. Also, the M-Series is an Oracle term. Fujitsu does not use it. So when oracle talks about having a T-Series and a M-Series merge into a common SPARC platform, that tells me that Oracle will be designing the next M-Series processor, not Fujitsu.
Hand Ellison right after Steve Jobs
Hahaha. Trying to become another Apple. Of course monopoly power is profitable...but as long as some countries favor other competitors or God forbid actually foster competition, Oracle will be like Microsoft and see it's Google or its Linux lol.
Itanium & Intel
HP partnered w/ intel as they could not afford to stay in the fab game. The design was good, apart from two problems
-dynamic compilation languages took off, it's hard to compile down for something as complex as PA-WideWord
-AMD pushed Intel hard, first on x86, then led the way on x86-64, removing the edge of Itanium performance.
I don't see Sparc or Itanium surviving for more than 5 years. Apart from the "big database" deals, x86 are it, in tens to hundreds. Yes, the hardware is commoditised, but it is nice that it's the same on your laptop as your server: no surprises. Yes, the OS is a "commodity" Linux, but it is nice that it's the same on your laptop as the server: less surprises.
Linux has been transformational, and helped kill Sun: why pay for Solaris on Sparc when you can get Linux on x86 -especially as that is the platform that OSS apps are being written for, that you can pay by the hour for on EC2?
What it has done for HP is removed differentiation between them, Dell, SGI in the server deals: the CPU and OS may have been commoditised, but that doesn't mean the layers above need to be. Which is where Larry makes his cash. Today.
ARM... think about it
you pay, they deliver
This sounds like good engineering. If they deliver significant performance increases with their HW/SW co-design, then surely plenty of people will be ready to pay for a better product. I cannot see what is wrong with designing something better ans selling it for a premium.
Database Software optimised for Hardware? Anyone smell shades of Netezza?
Larry Elison wants to buy everything in sight...
"Oracle CEO Larry Ellison reiterated his interest in buying a chip maker, but AMD CEO Dirk Meyer said his company is not for sale, though he would listen to offers."
Three months later Dirk Meyer was fired.
The year is not over yet. Larry could now buy AMD pretty dam cheap.
Larry only plays if he can be #1
Larry isn't going to play in any business unless he can be #1 so I doubt buying AMD is in the cards. Right now, my bet is on Oracle buying a serious storage vendor like EMC or Netapp. Then he truly has the entire stack and can control it all.
Oracle and T4...
I always enjoyed the Sun publicity mechanism. It was hopeful and told of things that the market appears to need. They smile nicely, too. It appears to be reawakening at Oracle, after years of low-use.
Delivery hasn't happened yet for T4. It would be great for Oracle to put pressure on IBM and Intel to generate higher-performance database platforms.