| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Usenet -> c.d.o.server -> Re: Snapshot log overhead
dpurrington_at_my-deja.com wrote:
>
> Thanks for the reply. Actually, what I'd like to know is what is the
> overhead involved with performing an insert during any write
> transaction. My understanding is that when you update a record, Oracle
> records the rowid of the changed record and the timestamp of the new
> transaction. There has to be some overhead for that transaction.
>
> You wrote:
> > I think your question though is really more pointed toward when does
> > trigger based replication become an overhead compared to log based
> > replication, like SharePlex or Oracle's own standby database. That is
>
> What I'm comparing it to is the Shareplex product. According to Quest,
> this is NOT a trigger-based replication mechanism. It monitors the
> redo logs and passes the information to the replicated system.
> Therefore, it has no interaction or impact on the monitored RDBMS
> (aside from the indirect impact via the CPU cycles necessary to monitor
> the log file, queue it, etc.). It does have some functional drawbacks
> though (you can't replicate horizontally).
>
> But both solutions are the incremental or "fast" refresh type of
> replications.
>
> I hope this clarifies the question! Thanks for your help.
>
> Dave
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
I did some timing tests (some very simple ones, ie, insert stacks of rows, then do the same again with a snapshot log)...
In 7.3, overhead was about 8%
In 8.0, overhead was about 3%
The main difference was that in 7.3 it was running a PL/SQL trigger whereas in 8 that trigger has been internalised.
HTH
--
"Some days you're the pigeon, and some days you're the statue." Received on Mon Nov 22 1999 - 03:13:49 CST
 |
 |