Oracle development on Itanium - and now ?

[Wim Rombauts]

There have been a few (now lengthy) threads on this forum about this topic. A lot of (bold ?) statements have been made.

But now that the first dust has settled a bit, what's up ? What do you think will really happen with your HP servers running Oracle ?

TMHO ... I liked working with HP-UX a lot. I find that HP Integrity hardware is no longer that much more expensive compared to equivalent x86 hardware. And for rather small systems, HP-UX support isn't multiple times that of Linux anymore. HP-UX had still that edge of advantage in rolling upgrades and migrations, that I have not yet seen in Linux (although I may not be well enough informed there).

I really believed in a future for HP-UX.

But now not anymore, because we (and I guess you) have mainly 3 options :
1) Stay with Oracle RDBMS, for whatever reason. Then that will be without HP-UX, and that will leave no software on HP-UX that makes it worth to keep that platform.
2) Move to DB2 or PostgreSQL Advanced Server, but have you looked at 3rd-party application support for these databases on HP-UX. It's a lot more limited than for Oracle, unless we run DB2 on AIX or Linux, or PostgrSQL on Linux.
And, if we need to move to a new database, there aren't that many reasons to stay on HP-UX. Choices will probably be maid to run that on Linux.

Unless ... HP manages to ramp up 3rd party application support for DB2 and PostgreSQL on HP-UX and make it a much more profitable solution compared to running the same on AIX or Linux, but I don't see that happening.

So I feel that Oracle is important to HP-UX, and with Oracle gone, a large percentage of current HP-UX customers will see themselves forced to leave the platform, possibly enough to make it loose critical mass and no longer really profitable. And with a forecast of dropping sales, Intel may be rethinking it's Itanium future, wich will only add to the death of HP-UX.

I feel sad about this. And you ? What future do yo see in your company, now that the dust has settled ?

[Michael Steele_2 · Accepted solution]

This problem was created when they decided to go with Itanium: An idea developed by marketers and salesman with architecture so similar to Intel that it becomes pointless to support both because they are so nearly identical. But salesman wanted a piece of the bigger Intel pond and also wanted a copyright. In the beginning of Itanium, Intel gave HP something like a 3 or 4 year copyright agreement. In the fifth year Itanium reverted back to Intel copyright.

I have to believe that HP had the option to continue with the first agreement. It certainly is preferable to have copyrights. I also have to believe that HP chose not to continue in the fifth year(?) with the old agreement and to allow the rights to slip away.


They also got the Compaq line of servers in 2000 â 2001. And the Proliant DL series (* formerly Compaq *) is probably HP's current biggest seller. But it runs on Intel, not Itanium so it supports LINUX and not HPUX 11.23â ¦

And this is the Oracle market: Intel on Proliant Oracle RAC on Linux.

And the switch over is going very fast. Itâ s a dramatic switchover going on amongst HP clients.

Most clients that I have talked to have gone in this direction. There reasons vary. Some are very emphatic about losing Pa-Risc and can't take the step. It's almost an emotional decision. More have been given this alternative by HP Sales and go with Proliant hardware and RHEL O/S support. But in this current scenario, HP only keeps a smaller sale of a server and the HW support agreement. They lose the O/S support agreement.

This has to be noticed. And is part of the reason for such poor O/S support which everybody complains about. Inflation probably being the bigger reason.

In my opinion going with this path only creates huge problems in the loss of UNIX STANDARDS between version of O/S's and applications for everything LINUX related is a huge, huge problem. There is no consistency or reliability in many Linux utilities between versions. Right now for example I'm changing the root login procedure for ssh in about 250 Linux boxes which range from version 3 to 4 to 5. And many cannot accept the change. Worse many cannot support an upgrade. Worse they have custom fixes by people whoâ ve left and therefore canâ t be maintained.

In Linux often times there is no recommended patch because a Linux hobbyist has not yet gotten to it. In the GNU market you have to wait for the volunteer supporting organization to get around to it. To get around this problem many clients write the code themselves and rename their LINUX O/S to something else like, Cisco Enterprise Linux.

But this too is a huge problem. Why? Because only perform minimal testing is performed in the one data center environment and you lose the conformity that you once had with copyrighted O/Sâ s like HPUX. And the learning curve becomes even greater for the new admin coming into a new LINUX data center.

[Michael Steele_2]

To underscore the similarity between Intel and Itanium CPU's, so similar in my opinion that its insane not to eliminate on the way Oracle has done, here is a common command that can be used by both Intel and Itanium: CPUID. Used to define CPU versions;

* 486 (1989): family 4
* Pentium (1993): family 5
* Pentium Pro (1995): family 6, models 0 and 1
* Pentium 2 (1997): family 6, models 3, 5 and 6
* Pentium 3 (2000): family 6, models 7, 8, 10, 11
* Itanium (2001): family 7
* Pentium 4 (2000): family 15/0
* Itanium 2 (2002): family 15/1 and 15/2
* Pentium M (2003): family 6, models 9 and 13
* Core (2006): family 6, model 14
* Core 2 (2006): family 6, model 15

Intel considers Itanium so closely related to the Intel CPU that they describe their difference only by different versions of the same architectural advances.

[Wim Rombauts]

I had hoped HP would have made a clearer statement.

I don't really need Oracle. I need a well supported database, and a platform that I can trust to be here for a long time. A platform that is worth running because of it's functionality, reliability, price, software support, ...
And a platform supplier that I can trust and depend upon.

But latest news from Intel (engineers moved to Xeon) and HP (nothing really) is not the kind of news that gives me confidence in the future of HP-UX/Itanium.

I know I can easily run Oracle 11g R2 on HP-UX 11i v3 up to 2020 (and that's a really long time in IT), and I don't really have to worry right now, but that's not how my manager thinks. He wants to know now in what amount HP-UX is a platform that is worth any further money.

So, HP, come out with a statement, a plan, a roadmap that shows HP-UX has a bright future, that HP-UX will even be better now that Oracle has gone, because. Let customers come first, like you claim, and give us confidence in your platform.

[Michael Leu]

IMO this is purely a anti-competitve move by Oracle in the Unix market and the talk about the demise of Itanium (i.e. HP-UX) pure FUD.

There are the public roadmaps from Intel on the next few generations, so that should be covered. Just the next chip should be a huge improvement on Tukwila


And in the very long run I guess Itanium will just be a Xeon chip with a different firmware and bigger margins. So I don't have too much doubt about HP-UX's future, and on what chip exactly I don't really care that much.

[Conor O'Mahony]

A couple of interesting surveys show where people's heads might be:

- IDEAS surveyed 66 people in these forums in the immediate aftermath of the announcement... http://ideasint.blogs.com/ideasinsights/2011/03/hp-ux-at-crossroads-as-oracle-ceases-development-for-itanium.html

- Gabriel consulting more recently surveyed 450 people... see the embedded PDF here... http://gabrielconsultinggroup.com/gcg-news-and-views/20-general-blog/272-survey-says-why-oracle-dropped-itanium.html

[Conor O'Mahony]

By the way, here's where I stand on this...

Disclaimer: I'm an employee, working in their data management software business.

IBM DB2 will continue to treat HP-UX on Itanium as a tier one platform. In the IBM Toronto Lab, the DB2 development organization continues to build the next release of DB2 in support of the HP-UX on Itanium platform and has no plans for any change in that platform's tier one status.

In fact, if you look at the history of DB2 support for Itanium, IBM has delivered HP-UX Itanium support on the same day that we have delivered new releases on AIX, Windows, Solaris, and Linux.

You can read more on my blog... http://db2news.wordpress.com/2011/03/27/ibm-db2-welcomes-oracle-databasehp-itanium-customers/

[Hein van den Heuvel]

In the early 2000, while working at HP, I had the pleasure of working with the IBM - DB2 team, notably in the context of SAP benchmarks but also for the product placement and strategies in general.

There was a good bit of management suspiciousness and pressure against that.
1) we wanted to play nice with Oracle.

See where that got them!

2) the fear that once a customer was on DB2, the move to AIX/PowerPC would not be far.

I never saw the latter as a concrete threat. The DB2 sales folks supposedly sold software only and were measure on software only. If the customer were to be converted / coerces to Aix, then the DB2 sales person would loose out. That was the message and it seemed to hold.

While during our technical work AIX certainly was referred to, it was always in a constructive fashion: "Here is how we do this on AIX, how can we get there using HPUX or is there a still better plan?"

I thought is was a promising relationship at the time with good commercial and technical potential, and I still believe that to be the case.... watching from the sidelines only.

fwiw,
Hein van den Heuvel

[Conor O'Mahony]

Speaking of SAP... a lot of SAP customers run SAP on DB2/HP. Its quite a common combo. And, as far as I am aware, SAP runs its own internal SAP systems on DB2/HP.

[Bill Hassell]

One area that I have not heard any comments from is Sybase sales and marketing. To me, it would seem to be an amazing opportunity to provide an enterprise database on enterprise hardware. Where's Sybase in the conversation?