So why did Oracle drop now? If you recall, when HP gave Mark Hurd, its former CEO the boot for things still under court seal, he popped up at Oracle before most people blinked. There still seems to be a lot of unsettled issues and even more bad blood between HP and Hurd. When his new company decides to pull probably the most widely used high end database on the market from the top of HP’s compute stack, Hurd knew exactly what would happen.
His reputation is a penny squeezing bean counter who lets nothing happen on his watch without knowing. If he didn’t know, to the penny, the effect this would have on HP, I would be shocked. The Itanium boxes are key to HP’s high end aspirations, much of it’s proprietary software stack, several of it’s OSes, and most lucrative, it’s support and services arm. The lines that HP sells with Itanium are so profitable it is silly once you take the ecosystem in to account, very similar to IBM and their Power chips. The lock-in is as extreme as the profits, enough so that HP could probably give away these huge machines and still rake in massive money.
If you think Oracle’s move is anything more than a naked poke in the eye toward HP, you are deluded. The is where the payback begins, but it is also quite plausibly deniable. One of the most obvious reasons is OS support. Oracle supports three main OSes, Linux, Windows, and now Solaris. Of the three, they own 1.5 of them outright, and the other 1.5 are competitors.
Since the purchase of Sun a few months ago, Oracle now owns Solaris, and that is now the OS of choice for running Oracle software. Windows is of course the opposite, owned by Microsoft and anyone who uses it for mission critical work, and runs the kind of software that Oracle sells, is, in this author’s opinion, downright crazy. Microsoft has shown a complete unwillingness, not inability, to secure their OS, so keeping the family jewels on it is simply asking for problems. Then there is stability, or lack thereof, but if you run anything Microsoft, you are probably painfully aware of both issues.
That is one own and one not, the .5s are Linux. Oracle has their own version of Linux, and also supports several others for a fee. Oracle’s distro is essentially Redhat with the logos and anything copyrighted buffed off hastily, then repackaged with their logos and some tweaks. It is basically Oracle’s way of taking the work Redhat did and getting paid for it. There is nothing illegal about this, it is quite permissible under the licenses, GPLx and others, that Redhat releases under, but ethically it is completely bankrupt. Redhat is now waging a tit for tat war with Oracle, making their life harder here and there where they can.
In short, Oracle loves Solaris because they own it. Redhat and Microsoft are enemies with competing products, so why not take a few potshots at them in a release that will be widely read and discussed? They did. I grinned. Most people didn’t get the joke.
Going back to hardware, Oracle got control of the Sparc architecture when they purchased Sun. Sparc has been waffling for a long time, it is far too little, too late, and not competitive with x86, much less Power or Itanium. On CPU performance, there is little to no reason to pick a Sparc based box over any of the competition. That leaves software. The only OSes that run on Sparc machines are Solaris and Linux, both of which Oracle owns or or controls. The other potential players, Redhat and Microsoft, bailed out of Itanium long ago. Not much can be done by Oracle on the OS level.
Once you get above that point, the the application stacks, middleware layer, and even the languages that are used to run the huge boxes that Itanium chips power are provided by Oracle. Oracle is not the only player here, but they are in a large percentage of the shops running Itanium, and likely more critical to the corporate IT ecosystem there than the hardware. The Oracle software stack likely costs more than the hardware that it runs on, so if you chose to replace one strictly on price, bye-bye HP/Itanium.
What you are seeing is a calculated move by Oracle to hamstring Itanium and boost their own Sparc machines, that much is clear. The obvious calculus is that the pain of moving away from Oracle is higher than moving away from HP/Itanium to Sparc, Power, x86, or god forbid, one of those and another software stack. Itanium was the low hanging fruit, and is pretty obviously a test case to dip their toes in to before going after IBM, Power, and other software vendors.
The fact that Oracle has Mark Hurd in a top position didn’t hurt anything either. He knows how loyal these customers are, and where the weaknesses are. He knows if they would defect if given the choice on Oracle vs Itanium/HP Hardware. He knows where the soft spots are. The fact that Oracle just put a knife in most of them? Sheer coincidence, there is probably a piece of paper somewhere saying he can’t do that, or use that knowledge after he leaves, and would he ever do anything like that? He as a CEO after all, if you can’t take his word for it……. To top it off, Oracle has had more than enough time to commission surveys about customer loyalty to provide any data needed to back up the decision.
Where do things go from here? You will see the Sun remnants, IBM, Microsoft, and every database vendor offering sweet migration deals for people to move away from HP, Oracle, and everyone else. Some business will be migrated away from Oracle and more from HP, but when the dust settles, the majority will probably stick with Oracle. Then the skies darken with lawyers riding on the backs of their winged monkeys. They are the true winners of this affair.
You might recall that Intel was the ‘official’ target of Oracle’s action, but they are just roadkill. Oracle probably doesn’t care much about Itanium, it was an easy target with lots of data available, a convenient test case. Use it or lose it, and Oracle tends not to dither around with things like this. HP had their eye poked too, hard enough for Intel to feel it and wince.
The response from Intel was swift and tepid, and contained not much more than the usual Itanium battle cry, “Wait for the next generation, or maybe the one after”. It has served Intel well for the past 5+ years, so why change the strategy now? They publicly hugged their partner HP, and the HP-UX OS, and promised to support everyone and everything that even thinks about Itanium. All seven might get a second card this year telling them that they are loved, and what more does an IT manager need?S|A
Note: Just before publication, we received word that HP has chimed in with an equally delightful missive entitled, “HP Supports Customers Despite Oracle’s Anti-customer Actions“. It pulls no punches describing Oracle’s actions as ‘Anti-customer’, and point out how the Sun, now Oracle, hardware is basically tanking in marketshare. While the press release wins no medals for being subtle or shining light on things not currently blindingly obvious, it is still a fun read.
The summary is that Oracle is bad, Oracle’s hardware is tanking in the market, HP supports their customer, and loves Intel, Itanium, and probably puppies too. HP’s CEO Leo Apotheker thinks the Intel cards manage to toe the fine line that is both spiffy and tasteful, and customers have nothing to worry about. They are not being gutted by an insider, and Oracle is bad, and HP does really love puppies.S|A
Latest posts by Charlie Demerjian (see all)
- SemiAccurate has Skylake-SP die shots, sizes, and more - Jul 21, 2017
- Skylake-SP has a diverged core - Jul 19, 2017
- Intel’s Purley platform architecture is a step forward - Jul 13, 2017
- Intel launches Purley aka metal Xeons - Jul 11, 2017
- AMD’s Epyc has lots of connections - Jun 26, 2017