Re: OEM 10g architecture question

dr a écrit :
> Can anyone point me to suggested implementation strategies for 10g OEM?
>
> specifically, where the different components have been intalled? for
> ex:, has anyone installed the management service on the same node as
> the DB? what types of performance hit that makes.
>
> Thanks in advanced!
>

1° I suppose that by "the DB" you mean the repository DB. Cause the full OEM for just one DB to manage : it's a little bit too rich.

2° On the same node, I did it with OEM 9.2 on a poor Mandrake Linux 9.1 PIII-392Mb : OMS + DB for repository. This repository addresses 50 DB on different nodes for a few BDAs. My experience is the DB for OEM repository is a litte one and is not often in intensive activity. It's running fine for 4 years. Sometimes (< 10 in 4 years) stop-start to kill defunkt JRE processes. In 10g the OMS is bigger cause of the web server. Just respect Oracle prerequisits, but I think repository DB won't be much more stressed.

3° "On the same node" is a way to have the whole administration stuff on one point : no network problem between components, independance for start-stop, "exp" sufficient for save those fews Mb,... "On different nodes" is a way to save the cost (CPU memory disk) of a DB by hosting the repository in a server running Oracle DB's. The risk is your repository is in a target DB : it's obvious but if you shutdwon that target DB with OEM, your OEM will hang.

Hope it helps Received on Sun Apr 23 2006 - 12:32:47 CDT