As with any solution, the more frequently you poll the less you will be able to poll. The polling rate is indicting that you are reaching the upper limit of what a single polling engine can handle. I would recommend leveraging passive monitoring wherever possible to lessen your reliance on active monitoring. Utilize SNMP Traps, syslog, and the Windows Event Log Forwarder wherever possible, as these methods will notify you instantly when an error condition has occurred so there' no need to wait until the next poll cycle. Another option is to break those components within a template that are most critical to monitor (processes/services/UX monitors) out into their own template and poll those items at a higher rate than items such as performance counters etc. This too will lessen the polling burden at the upfront cost of creating and customizing application templates. The easiest option is to add an additional poller to increase polling capacity and spread the load out so you can monitor at your desired 1 minute interval.
↧
