Re: Oracle8 parallel server ???????

I think that a few days is far too short a time to do an OPS install on a brand new set of hardware (even if it is certified). OPS is very fickle on hardware and software patch levels and clustering patch levels, etc. On a DEC unix system I had we spent a while debugging and looking at core dumps and inconsistencies until we eventually found that both the TruCluster and OS patches needed an upgrade.

You wind up in the middle of that common problem of vendors pointing fingers at each other claiming that their side is fine.

In short, these types of strange problems should be counted on for any project that is cutting edge and new to you.

Billy Verreynne <vslabs_at_onwe.co.za> wrote:
: Greg C. wrote in message <01bdfb77$df981d10$360baa0a_at_ictltgclark>...

:>What are the Oracle requirements for implementing OPS on a RS6000 interms
:>of hardware, personel roles and time. I recently heard that a shop required
:>1 month to configure their HP machines for OPS. Can anyone tell me what
:>kind of time line I should expect to implement OPS?  Are there any gotchas?

: 1 month!! We configured an OPS on a Unix-based MPP recently in a couple of
: hours. What takes time is to prepare the machines. Installing Oracle,
: configuring raw devices, installing and configuring the DLM (distributed
: lock manager) and whatever else needs to be there from the o/s side. But
: this should take maybe 2 or 3 days at the most. Quite likely it was a first
: for whoever did the installation you described.

: Creating an OPS database is very easy. No gotchas. Just remember to make
: sure each instance has its private rollback segments and to enable redo log
: threads. The OPS manual (chapter 15) explains it very well.

: Oh yes, we also used a global init file for the global db settings and a
: local instance specific init file that includes the global file.

: regards,
: Billy
Received on Fri Oct 30 1998 - 18:56:20 CST