Is there anyone that does Librenms consulting? Our install is stuggling with the scale of our network and wanted someone to look at and see if there is anything that can be done to help the matter.
I would be very interested to know the resources you currently have supporting all those devices and ports, and the outcome you get to better the performance. I have 486 devices and 8127 ports and already its heavy on resources.
Right now it is vm on a proxmox host. The host has dual Gold 5318N.
The VM has 60 core with 2 sockets with 30 core each and 32 gb ram. Disk is on a pair of nvme disks in a mirror zfs array.
I am thinking of getting one of the new x70 dell servers with dual Xeon 6 Efficient 6740E 2.4G, 96C/96T,
It does not help that 500+ of our devices are 48 port cisco swtiches and they take a long time to poll and a few have a few site that are connected with vpn over starlink while we wait on fiber to be installed. Those ciscos take an avg of 210 sec to poll each.
yeah nice, other than the documented optimising suggestions i found that using mysqltuner and increasing the snmp flow-control pps on each 48 port switch cut the poll times by around 40 seconds alone. in total our poll times are at 70 seconds but i can bring that down to 60 if i throw more cores at it.
Do they drop if you up max repeaters on them? Depends on network and devices, but my sweet spot on bulky Cisco gear has been 70 lately. Makes a massive difference, particularly if any latency is involved.