[Pgcluster-general] speed of updates

[Christian Dannemann]

Hi All,

I've done some testing and discovered that the speed of updates in in 
straight relation to the number of databases in the cluster.

This makes me suspect that updates aren't done through different threads 
at the same time, but sequential - one after the other.

If the latter is the case, I would consider that as a bug in the code. 
The behaviour I would expect is as follows:

- update/insert/delete query comes into one node of the cluster
- pgreplicate detects that this is a db change and starts a thread per 
database to execute the update
- after all n threads have come back with success, the updates are 
committed across all n databases
- next update in line is treated in the same way ...

Could somone confirm or alleviate my worst fears?

Thanks,

Christian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://pgfoundry.org/pipermail/pgcluster-general/attachments/20070215/8657d074/attachment-0001.html