Latency Issue

[sush_1]

Hi Team,

• We are facing increase in the latency when task is getting stuck at where it is waiting to commit transaction at the source .
• after 1 0r 2 hrs task will start running and latency will come down and  it will wait for transactions to commit at source side. and same thing will happen again and again.
• Why the environment is behaving like this ? Is this issue with replicate side or something else?

[john_wang]

Hello @sush_1 ,

Thanks for reaching out!

This is normal behavior if you are run huge size changes in a single transaction in source side Oracle database. I'm assuming the process is: begin a transaction; update many number of rows; commit the transaction. and repeat the similar operations in source DB.

From Replicate perspective, Replicate detects the changes from REDO Log, then start reading the changed rows from source DB and repeat the read until get COMMIT or ROLLBACK flag so far Replicate knows how to handle the transaction: commit to target, or discard the whole transaction. Before Replicate get the COMMIT/ROLLBACK, the latency builds up and you see latency is increasing in GUI, until it drop down to 0 sometimes.

If you want to get smaller latency, then splitting the huge number changes into multiple smaller transaction may help.

Regards,

John.