August 3, 2017 at 9:10 pm
Hi Community
Does anybody have any experience using replication perfmon counters to track replication latency ?
We have transaction replication in our environment and there are some alerts set up which use replication perfmon counters like latency, replicated cmds / sec to track latency. I have often seen the values reported by these counters are not inline with what we see in replication monitor or number of undistributed commands. Even setting up tracer tokens reveal that replication is indeed healthy. Still the counters show way too high value for latency. Also the counter type is 65792 which means it just gives the polled value and no further calculation is required.
Why is this discrepancy in what perfmon counters show and what the actual latency is ?
This makes me believe that are these counter values bogus and far from true ?
If any body can throw some light that'd be great and also how can we test actual latency using perfmon counters ?
August 4, 2017 at 4:21 pm
I'm not familiar with the perfmon counters, but I have seen wonky reporting in Replication Monitor. Several times I've seen it list a Publication's performance as Excellent, when it actually had over half an hour of latency. We've joked "Excellent" really means the publication is "Excellently Latent."
Viewing 2 posts - 1 through 1 (of 1 total)
You must be logged in to reply to this topic. Login to reply