Oracle Ends Partnership With HP

Rambo Tribble writes "As detailed in a Reuters report, Oracle is terminating their cooperative relationship with HP in light of their anticipated acquisition of Sun. With Sun servers in house, Oracle apparently feels no need to work with HP anymore. They will 'continue to sell the Exadata computers, built in partnership with HP, until existing inventory is sold out, if customers request that model.' Oracle is much more enthusiastic about a new version of Exadata, which they developed with Sun."

Comments

motivation for purchase

[uncreativeslashnick]

This causes me to speculate if the reason behind the purchase of sun was that oracle didn't like doing business with HP, or saw that HP was making a ton of cash off the deal.

Re:motivation for purchase

[CharlyFoxtrot]

Maybe it's the other way around: Sun is sinking fast due to uncertainty and so they make some bold gestures to show they're serious about making this merger work. Nothing shows you're committed like eating your own dog food.

Lions and tigers and bears, Oh My!

[symbolset]

There's going to be a lot of shakeup over this one. IBM and Dell must be pondering the enduring fidelity of Oracle in a world where they make their own servers.

And that's a two-way street.

IBM

[Nick Driver]

What IBM needs to do now is make a new version of DB2 that's fully software-compatible with the Oracle API so that you can take an application that's written to run against an Oracle database, and have it be able to talk to a DB2 database without being able to tell it's a different brand of database engine.

A long time ago I worked with an outfit that made a translation layer that let an app that was written to run against an HP3000 Turbo Image database, be able to open up and read/write to an Informix database running on any Informix-supported platform anywhere on the network. The app had no idea it was talking to a different database, it was 100% transparent.

If IBM could do something like that for DB2 to emulate Oracle, they could greatly undercut Oracle's expensive stranglehold on the mid-sized market where customers already have CRM software apps that are written for Oracle databases and they can't upgrade to the newest multi-core processor hardware because Oracle's licensing costs are so expensive.

Re:IBM

[fsmunoz]

What IBM needs to do now is make a new version of DB2 that's fully software-compatible with the Oracle API

See here and here for example.

 

The Oracle compatibility feature will enable Oracle applications to run natively on DB2. In discussions with Gartner, reference customers tell us that DB2 runs 95% or more of Oracle-specific functionality found in SQL statements and natively runs PL/SQL, Oracle's stored procedure language. This is native functionality; it is not an emulator, nor does it require changes to the application code (other than the 5%, which is mostly minor functionality, not found in many applications).

Having said that, and while it is a worthy and very valuable feature, there is more than compatibility in play when trying to pitch a change in DB engine.

they can't upgrade to the newest multi-core processor hardware because Oracle's licensing costs are so expensive.

Not only that, but Oracle applies modifiers according to the processor type. This is in principle not something odd: it makes sense to differentiate per CPU type given the sometimes staggering difference in terms of processing capacity (IBM does the same with the PVU-based pricing). However, given the Oracle acquisition of Sun this could mean that Oracle will tilt the modifiers even more (last time I checked Sun cores had a 0.25 modifier value, the lowest of the lot).

Re:IBM

[drinkypoo]

I don't know if this is still true, but DB2 was the most scalable but slowest of the major RDBMSes last time I looked (on most hardware that would run all of the above.) Converting from Oracle to DB2 would have performance considerations even without a translation layer (not that Oracle was one of the fastest.) In order for this to have a hope, DB2 would have to have near-complete feature parity with Oracle. That's not impossible, but I also don't think it's true at this time.