latency due to high volume

[suvbin]

Hi Team,

If the latency is due to high volume of data, then how would you trouble shoot it or what are the optimizations we can perform on it. And what are the data validations scripts we have.  or how we can do the data validations.

Thanks

[Heinvandenheuvel]

First step is to determine whether it is source or target latency. PERFORMANCE TRACE logging may help, using a NULL target may help.

If source, you nee to know whether the source fails to deliver quickly enough, whether the network can handle the volume (you better know how many MB/sec is being requested), or whether Replicate SOURCE_CAPTUE does not have the cycles to process (CPU task/thread monitor tool).  You may also want use some TOP SQL query from Replicate-user on the source to help drill down.

If target, you typically want to figure out which table is the lead latency contributor. TARGET_APPLY TRACE (DEBUG) is your friend. Carefully study. Look in this forum for my (perl) script to help with that. You may also want use some TOP SQL query from Replicate-user on the target to help drill down.

Hein.

[SushilKumar · Accepted solution]

Hello @suvbin 

As you queries have its answers hidden in it.  AS Higher volumes of data also consume resources at database itself. And QR also took longer to process the changes.

As Hien mentioned it require analysis of participating components to know the cause of latency. 

As a Simple suggestion high volume of data generate high number of Log to read by QR if they are on Slow read Diks. you may see Source latency as well

If target is not able to apply the changes in the same way they supposed to do. Require analysis of Target DB as why DML operation are not applied. Reason may be Target DB resource issue or Something else.

Regards,

Sushil Kumar