net.tcp.service.perf timeout
For my appliance, it is important to follow TCP connectivity between certain hosts. For this, I have deployed a row of net.tcp.service.perf keys.
Graphing those, I get a lot of 3 sec spikes, indicating probably TCP timeout. Is there a way to to suppress those spikes?
Say, if usual value is between 20-200msec, than 3sec spikes make the graphs generally useless. So we need some kind of sanity check.
Is the only way writing procedures directly in database?
Thanks,
R.
Graphing those, I get a lot of 3 sec spikes, indicating probably TCP timeout. Is there a way to to suppress those spikes?
Say, if usual value is between 20-200msec, than 3sec spikes make the graphs generally useless. So we need some kind of sanity check.
Is the only way writing procedures directly in database?
Thanks,
R.