Oracle 10.7 on HPUX 11.0 - productivity hit

[CM-Corp]

I guess I finally have to take the hit and upgrade a 4way K570 to 11.0 for the Oracle beast. I've done enough HPUX upgrades in other environments to know that a clean install of HPUX 11.0 is the only reasonable (sane) way to go.

My real concern here is the productivity hit that I can expect. We plan to install split 8.1.7/7.3.4 which will give us some 64-bit support, but it also means that most of the 10.7 Oracle apps binaries will be running under 32-bit compatability mode. Upgrading 10.7 at this point is not an option.

So the main question... what kind of real world productivity hit can I expect? Is this a case where I can live with it, add a couple of processors, or buy stock in HP? Any user experiences will be greatly appreciated. BTW, it seems the many of the Forum threads on the topic have disappeared.

Tim

[Bill Hassell]

You won't see much change in going from 10.20 to 11.0 HP-UX. If you install the 64bit version of HP-UX, you can run both 32bit and 64bit instances of Oracle. It's not possible to predict what performance differences there will be. For instance, the buffer cache management in 11.0 is muchy better than 10.20 and so is networking, so it should be a bit faster for I/O.

The biggest change will be RAM. Start with 4Gb and see if your DBA can improve the 64bit Oracle version with a larger SGA. The 32bit version will still be as constrained as it was in 10.20. You can expand the 32bit SGA by using memory windows...check the white papers in /usr/share/doc on mem_mgt, proc_mgt and memory windows (mem-windows will need recent patches to show the white paper).