I’m looking at the Replication Guide and noticed restriction #3
"It is not possible to access the replication database during replication processing"
Now I wonder what this exactly means.
What I want to do ist use the replication database for read-only access.
Is that possible?
What exactly means “during replication processing”?
The way replication works is that log spaces from the master database are restored onto the replication database. For the application of the logspace the replica database is shutdown. So during this time the database is inaccessible.
Hope this helps.
Software AG (UK) Ltd.
thanks for your quick reply.
So I understand there are 2 ways to replicate:
1. Application of the logspace is batch-wise and has to be initiated manually.
2. I can set it up to automatically pull the logs constantly but then the replication database will be inaccessible all the time.
Is that correct?
If replication is set up to be initiated manually, is there a way to initiate it via a script so that the log space is applied - say every 5 minutes?
Thanks again for you help
By setting up a replication database, the application of the logspaces is automatically done every 5 seconds, so it is virtually inaccessible. The update rate defaults to 5 seconds but this can be changed. The parameter is called “replication delay”.
Software AG (UK) Ltd.
So if I set the “replication delay” to - say 1 minute and the update rate on the master is very low (not more than 3 small updates).
Do you have any experiences, estimates, etc. how long the replication processing may take?
If I submit a read request on the replication database during that time, is the request suspended or aborted?
The answer is a little bit different: with 4.1.4, you cannot query while the replication database is replicating (this feature will be available with the next version). So if to query replicated databases, shutdown the replicatiion database, restart it “read only”, the do queries. If you shutdown again and restart without “read only”, the replication database is brought to the most current state again (and cannot be queried in that period)