Why so slow to post this news?
I wondered why el Reg was taking so long to post this news when it has been out there for nearly a week... then I realised TPM has to go check with his marketing chums at IBM exactly what he had to write ;o)
Software giant and systems player Oracle has restarted software development for Itanium-based machines. That sound you hear is a sigh of relief from HP-UX system customers the world over and from Hewlett-Packard, whose Integrity and Superdome servers fell off a cliff, in terms of sales, after Oracle pulled the plug on Itanium …
Sounds like someone is a little unhappy that their favourite platform is dying.
TPM talking to IBM, Intel, HP or Sun won't make the world suddenly like big, expensive iron over cheaper options... Itanium and Sun are losing ground to x86 faster than IBM because IBM still has performance advantages.
As soon as HP provide x86 versions of NonStop and OpenVMS, Itanium will be officially dead rather than the current state where it smells dead, but the doctor still detects weak signs of life...
"....if those customers would have bought Power in the first place." Ah, but you forget to ask yourself, "if Power is so good then why did they buy Itanium?" We have some very good reasons for buying hp-ux on Itanium - availability of applications, performance, reliability, ease of use, good support, and usually a pricing advantage over IBM Power. I'm betting many of the other Itanium customers had exactly the same reasons. I'm guessing you didn't think of them becuase you simply don't use hp-ux or Itanium.
"Not sure exactly what HP is hoping to achieve with this...." Well, I'd say that's obvious - keep selling new Itanium systems to customers with Oracle databases. The other Oracle software is pretty much unimportant.
".... do they really think that Oracle are going to do anything other than lip service porting of the latest versions...." Actually, Oracle have to maintain the same levels of support and development activity. Any cutback would have to be justified otherwise hp could go back to court again.
"..... I wouldn't be surprised to see them actually crippling the code to make it slower on Itanium..." Yes, I'm sure there are a few daydreams like that in Armonk. The reality is that, despite the constant IBM and Oracle FUD, Itanium sales didn't die completely since the original Larry episode, and the return of Oracle support can only add to the ramp up with the new "Poulson" chips due very soon. IBM had obviously hoped hp would still be in court come the "Poulson" release, I bet their marketing teams are going to be burning the midnight oil as they are going to have to rewrite all the FUD they had prepared!
Aw, is ickle Phil 4 upset that his beloved Snoreacle got smacked down in court? I know the Great Snoreacle said that they would replace all those Itanium servers out there with shiny SPARC systems, are you now worried it means even less of a chance of anyone buying any SPARC servers? Have a glass of milk and a cookie and console yourself with the thought that at least Larry's stifling of MySQL will keep the Oracle DB revenues rollingin for a while longer. On x86, that is.
Working as oracle consultant I know situation in quite many companies and really I don't know anyone who would went for exadata because oracle tried to get rid of hp-ux. If anyone wanted to buy that, they did, but hp-ux situation didn't have any importance here. In all cases known to me exadata replaced old Sun iron, not HP. Those who were anxious of hp-ux future either went to IBM or, if they wanted to stay with HP, they went for proliant and linux, And it was most often scenario during last 2 years. HP-ux troubles increased high end proliant market share, which was probably still bad for HP margins, but not that bad as it could be.
I don't believe this court decision would make any new customer to buy into hp-ux, but it's definitely a relief for existing customers, giving them few more years of flexibility to choose what to do. Everybody knows that unix is dying, we don't need courts telling us that, that's only a matter of having enough time to decide what next. Especially that current economy doesn't help with making expensive platform decisions.
And about oracle products and especially oracle support quality for future hpux: well, has it ever been any good? I don't remember such times. It cannot be much worse than it already is.
TBH, I wish the issue had never arisen. I don't give a hoot about the rest of the Oracle software, but we are heavy users of the Oracle DB product (on hp-ux, AIX and Windows). I stated when Larry bought Sun that I hoped it wouldn't screw up the relationship with hp as it was something that would adversely affect us. Larry then went on to shoot himself in both feet. I know other users that echoed the same sentiment - "just don't breaK Oracle DB on hp-ux!"
Hilariously, it looks like hp-ux is now the only OS which is GUARANTEED to have Oracle DB available for the future! Sure, there is little chance of Larry ditching Oracle DB for Windows due to marketshare, and probably for Linux (at least Oracle Linux) for the meantime, but no other OS has any guarantees. Slowaris? Heck, Oracle don't even use it for their appliances! AIX? None at all! If Larry wants to go on a new offensive tomorrow then IBM do not have any contract like hp's to stop him saying he will cancel all support for Oracle products on AIX. IBM fanbois can claim IBM would just offer DB2 instead but we all know DB2 is just not as popular as Oracle DB. In short, all the FUD that IBM threw at hp-ux whilst the courtcase was being settled still applies to AIX. Maybe all the Oracle users should be looking at hp-ux for some guranteed availability in the longterm.
"....and a few proprietary mainframe platforms in Europe and Japan, have been ported to Itanium...." Oh dear, TPM is suffering from convenient memory loss yet again! Seems it just slipped his mind - accidentally, of course! - that one of the mainframe operating systems that was ported to Itanium was IBM's own by a company called Platform Solutions Inc.! PSI worked out an emulator that allowed IBM mainframe customers to run their IBM mainframe software much faster and more cheaply on hp Superdomes. Indeed, PSI represented such a threat to IBM's golden goose that IBM, threatened, sued, bought and stifled the company (http://www.theregister.co.uk/2008/07/02/ibm_buys_psi/).
IBM also bought and killed Transitive (http://www.theregister.co.uk/2008/11/25/ibm_transitive_options/) as that allowed ex-Sun customers to run SPARC-Slowaris apps without a recompile on Itanium systems (http://www.theregister.co.uk/2006/09/27/solaris_itanium/). It seems that IBM have taken the Itanium threat much more seriously than TPM remembers. This is my surprised face, honest.
Don't worry, TPM, even if you "forget" the Internet never does.
I think TPM is talking about all of that legacy Fujitsu stuff, not z. Number of z/OS ports to Itanium can be counted on one hand with some fingers left over. IBM took Itanic seriously, even brought out some System x servers based on it in the mid-00s. Who wouldn't have taken Itanic seriously based on the early 00s hype? Intel said it was going to be the chip for everything replacing x86... just never panned out.
Given how fat Oracle grew on the back of selling their SW to HP-UX sites you'll probably see a bump up in Oracle's numbers on this decision. Never forget just how big the HP-UX running Oracle market was. Oracle made more out of it that HP did and TPM has guestimated the losses to HP of the war, Oracle will have lost more, a lot more. The only winner out of the Oracle/HP war has been IBM (OK and RedHat) big HP sites became not only scared of what would happen to their current platform but what Oracle would do next, if for some strange reason you don't think Linux is ready for you big systems yet, then IBM are left as the only show in town.
As others have said, this won't win HP-UX any new customers, but it will make a lot of existing customers more comfortable that they can plan their migration at their own speed rather than staring down the barrel of gun.
As for Oracle crippling their code, as tempting as it might sound if they got caught doing this the court is likely to through a fit that would make the most rabbit of management rants look like a teddy bears picnic.
Even if Oracle supports Itanium as normal for the next decade it won't change the fact that there was 18 months of uncertainty for HP-UX, and people spending millions on hardware hate uncertainty - this is why "you can never get fired for buying IBM" and later "you can never get fired for buying Microsoft" became rules to live by for cautious IT managers.
There will still be people worried that Oracle won't give proper support on Itanium, or will try some other legal manuever down the road, or will dump support when the contract with HP expires (not sure when that is, but I presume it has to expire someday?) The people who already fled HP-UX or are well into their migration plans will not come back as a result of this. Oracle permanently damaged HP's HP-UX market. HP may be able to get damages from them but they probably won't amount to what it costs them in the long run.
It'll also make it much more difficult for HP to migrate HP-UX to x86, if they ever want to do that, or even to add HP-UX compatibility into Linux. They know Oracle will be waiting in the wings ready to cut off HP-UX on Itanium if they can find an opening, and make HP-UX customers more concerned about its future than they already were the past 18 months.
Personally I think this was a calculated move by Hurd and Ellison, who did it knowing that they couldn't abrogate their contract with HP and would probably end up owing them damages. They just hoped to keep its future uncertain for long enough that it was damaged enough it became unviable, and they'd pick up enough former HP-UX customers to offset what this costs them. The only real winner here was IBM, who picked up AIX and DB2 customers out of this mess without spending a dime!
Very true. For all enterprise apps there is a golden combination of hardware, OS that the app runs best on. The app makers marketers would have you believe that each platform is an equal ("of course you can run SAP on PowerPC!") , but that is always, 100% of the time, rubbish. It's the dirty little secret of cross platform development: sure the developers make sure that they cross-compile cleanly, but ask 'em what they use in their offfice to test things out on or what their support organization have in the back room, and you will start to uncover the truth.
Time was when Oracle's golden combo was with Solaris on SPARC. Sure, it'd run ok on AIX or HP-UX, but it wasn't *quite* the first class citizen that it was on Solaris. Then they got religion and made the golden combo Linux on x86. As did IBM.
Today, I can see very, very few reasons why anyone would run Oracle on anything other than Intel x86. It doesn't really matter if Itanium sells now or not, or if it runs on HP-UX, or AIX, or Solaris. The sweet spot for this app is on Linux/x86, case closed, move along.
> Today, I can see very, very few reasons why anyone would run Oracle on anything other than Intel x86.
Let me fix that for you:
Today, I can see very, very few reasons why anyone would run Oracle.
Any business that wants to make the sorts of dirty moves and sharp practices that Oracle have in the past few years won't get my custom. As has been said before, Oracle doesn't have "customers", it has "hostages". Those that claim to love it are suffering from Stockholm Syndrome.
".....The sweet spot for this app is on Linux/x86...." Whilst that could be argued, what can't be argued is that Snoreacle have also hung a dirty, great-big axe over Oracle-on-Linux. Their intent to drive all Linux Oracle customers to Oracle's own Linux is making many in the Linux community uncomfortable, especially when they see Larry and Red Hat facing off. Then you have Larry's determination to push his walled garden offerings like Exadata and suddenly the Oracle-on-Linux option starts to look even worse. Unfortunately, it looks like the best option is Linux, just not with any Oracle products in the stack.
As has been mentioned, Oracle DB is only officially supported for OVM on x86. You can get it to work with VMware, but Oracle will only respond to tickets if it is a known Oracle DB issue or if you can prove to Oracle that it must be an Oracle software issue... meaning you need to remove VMware and run bare metal with the same issue. If the issue is an Oracle and VMware integration issue, call VMware and see if they can figure it out without Oracle or buy an Oracle supported hypervisor. They are also going to get midieval with you on licensing with VMware. You are used to hard partitions on Unix... well, on VMware you are going to license every core on the server.
Yes, Oracle loves x86 - Linux, provided you are talking about Oracle Linux on Oracle VM. Most people are talking about RHEL on VMware, which Oracle does not like. Their licensing and support terms for VMware are pretty outrageous. Also, Oracle likes x86 because it allows them to sell RAC more often... not because their software works better of scale-out vs. scale-up or anything of that sort.
"Yes, Oracle loves x86 - Linux, provided you are talking about Oracle Linux on Oracle VM....." One of the problems for Oracle is that on x86 there are more options than on UNIX. For example, Oracle have gone out of their way to stifle the threat of MySQL. Red Hat's nucleur option is that if Larry ever gets too precious they can bundle in something like MariaDB with RHEL. The 800Lb gorilla is MS SQL, which is cheaper and integrates better in Windows environments, and Larry can't afford to upset Redmond too much because the reality is Windows is probably going to be a key battleground for Oracle in the next ten to fifteen years. We're keeping a close eye on Oracle's support for Hyper-V as a bellweather to Oracle's despertation in the Windows market.
This is the key statement courts are going to have to deal with eventually. Oracle aren't the first, and they won't be the last. MS did it to Netscape.* And I'm sure it happened before that as well. There are certain instances where what ought to be a symbiotic relationship between companies can be manipulated by one and only one of the partners to its advantage. And once the action is done, the other partner can't recover because in the two to three years it takes the case to move through the court system, the software market has moved through two to six generations of hardware and software.
*Whether or not you think Netscape was a GOOD implementation of wc3 standards or not, it was the dominant player until MS illegally leveraged their market, even if the courts couldn't quite understand the issue.
Biting the hand that feeds IT © 1998–2019