3 Replies Latest reply on Aug 16, 2012 10:15 AM by jhay317

    How is Polling Load Calculated?


      Does anyone know what metrics and calculations are used to arrive at the percentage polling rate on a polling engine?


      We have polling servers which aren't breaking a sweat but the engine polling rate is shown at 106%. Something other than CPU/Memory must be driving this measure eg. a software or process limit.


      This is important for us because we have a large estate of 4000+ nodes and we're planning to roll out a package of custom UnDP statistics collections. We don't want polling intervals to be adjusted automatically and we don't want to over-provision polling servers if this isn't the bottleneck.


      We need the detail behind the polling rate calculation so we can understand specifically which variable has the most effect eg. node/interface element counts, status/statistics polling rates, UnDP MIB object sizes. We'd also like to know how this calculation is related to Polling Completion % and Total Job Weights and understand how these figures are derived.


      Any help appreciated!