Polling on 12.2 vs 11.5

Question

I recently went from 11.5 to 12.2 on my NPM.  I used to poll 1,780 nodes split up between two pollers as follows.

Poller 1  - 1,160 nodes - 49% polling maximum polling rate

Poller 2 - 620 nodes - 48% polling maximum polling rate

I now have 12.2 with 965 nodes

Poller 1 - 340 nodes - 54% polling maximum polling rate

Poller 2 - 625 nodes - 82% polling maximum polling rate

Note, the the new NPM 12.2 is on newer server hardware with better CPU and memory than the old server.  But if these stats are accurate, I won't be able to monitor the same number of nodes in the new environment.  I'll never be able to add an additional 815 more nodes.

Has anyone else had similar experiences?

Michael

mesverrum · Answer

Its helpful to have the more specific breakdown of nodes/interfaces/volumes as each of them has a different polling effect

Did you load up those queries to verify the intervals between everything is the same, including any one-off modifications?

There weren't any dramatic changes to the way polling load was calculated, and it actually has no relationship at all to your server resources (for example more and faster cpus will not reduce the load%) so if polling load is reporting higher in the GUI then one way or another there has been a change to the number and mixture of elements and the frequency at which they are polled.  Where resources come into play is the completion rate, as in a server without enough CPU to perform the scheduled polling will show a lower completion rate as things start to fall behind.

mkocsis1 · Answer

I also extended some of my polling intervals to try to mitigate this.

On about 542 nodes,

node status 240 seconds (default is 120)

Node statistics 15 minutes (default is 10)

mkocsis1 · Answer

Old environment

Poller 1 - 6,285 elements

Poller 2 - 18,015 elements

total 24,300

new environment

poller 1 - 6,743 elements

poller 2 - 10,242 elements

total 16.985