Collector performance

Generally the collectors are at idle and does not use much resources on the host machines they run on. The one exception is when a new log is added to monitor which requires a bulk import - this can take some time (minutes) depending on the remote machine, network speed and local resources.

Flood situation

One other potential problem is if there is a 'flood' situation in the event log of a machine being monitored where some application is writing dozen or hundreds of entries per second to the log. That is of course not only a problem for Event scavenger but for the whole environment. Event scavenger has something built in to try to handle this by avoiding writing too many similar entries to the database - a repeat counter per entry that gets increased for similar entries within a configurable time frame (default is 2 seconds).

Performance counters

The collector services have built-in performance counters to monitor its performance. The following counters are available:
  • Active event logs
  • Database accesses/Sec
  • Errors/Sec
  • Event entries/Sec
  • Total event logs

Viewer

The viewer application is written so that by default it limits the number of entries returned to the user. There is a way to override this (with an unavoidable 'warning').

Last edited May 3, 2010 at 8:32 AM by RudolfHenning, version 1

Comments

No comments yet.