Re: Multi Master Replication Slow

Set your delay_seconds up to something like 20 seconds, to make sure the push job is ALWAYS looking for pushes to be done.

• Make sure "min_communication" is turned on, if you are 8.0 or higher.
• Make use of column groups, so that only data from a given column group is sent over the network, when a member of the group is changed... as opposed to the whole row of data.

Other parameters to look at:

- JOB_QUEUE_PROCESSES
- PARALLEL_MIN_SERVERS
- PARALLEL_MAX_SERVERS

There are not too many differences to tuning replication vs tuning a standalone instance, so most other tuning guidelines apply.

HTH. Tom Best

<danthehorn_at_my-deja.com> wrote in message news:8je0ig$491$1_at_nnrp1.deja.com...
> I have setup three hosts with multimaster replication. Everything
> seems to work O.K. but it is very slow.
>
> I am selecting a count from a join on def$_aqcall and def$_destination
> to monitor each host since the purge is delayed. I am only able to
> push 600 transactions a minute to each host.
>
> I have setup 10 second pushes with parallism => 4 and on minute purges.
> I have moved the advance queueing tables off the system tablespace
> but nothing seems to improve the speed of the replication. ASYN
> replication.
>
> Any advice
>
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
Received on Thu Jun 29 2000 - 00:00:00 CEST