Hi Ivan,
it is very kind of you to reply so quickly. Are you working 24/7?
To tell the truth I've made a mistake. The number of new records per minute
is only 180 - and not 8100. Sorry.
Well, the replicator is ( actually was, because it had been stopped )
started every one minute. At the begining (last november) it took the
replicator a couple of seconds to complete its task.
As I can remember the odbc queries are always embedded in a separate
transaction. So I don't use any additional transactions. Maybe I make a
mistake.
But I'm sure that the collection of the data from the Oracle is very fast.
The speed of the communication is also ok, because the synchronized data can
be quickly queried via the web (intranet, 100Mbs).
The problem must lie somewhere in the pg or in the odbc driver. The rest of
the time must be lost somewhere there.
I also use a few indexes on the tables and they have to be updated 180 times
every minute. Could it be too huge a load for the server?
reguards,
Csaba
> As begin try to execute the replicator program in 5 or 10 min.
> Are you using any transactions by data inserting?
> How long take it to collect the data from oracle( it is extrem fast by
normal
> work)?
> How long take it to insert this data in pg?
>