LibreNms poller Performance metric

We are currently using Cacti with 53k data sources. The poller runs every 1 min, and it completes within 40 sec…
we are planning to move to LibreNms, Does LibreNms poller will be able to handle this(around 400 Devices) with single poller?
My system is 48Gb RAM and 16Core system with SSD drives.
I wont be able to run remote pollers or any additional pollers due to less resources…

i couldn’t find any poller benchmarks page available in Internet.

Also is it possible to config Librenms to use 1min poller for certain graphs and 5 min poller for certain graphs?

Specs should be fine.

The server doesn’t normally have much input into polling time, its normally network latency, how busy the end device is, and the amount of data collected. You can turn off poller modules you don’t need to decrease the time polled.

The polling time will probably be longer on librenms as it provides more information by default. I would test this with one of your devices first.

Unfortunately Librenms cannot currently mix and mash 1 minute + 5 minute polling.
See 1 minute and 5 minute polling on the same server

Poller comparison between Cacti’s spine and LibreNMS poller maybe very interesting information imo for both communities.

Based on resources it’s overkill; based on implementation it’s a disaster.

You should convert that machine to a virtual server manager. Create VMs to distribute work. Make a web access, polling nodes(set one for fast) and maybe a dedicated discovery/service check nodes.

With a more modular setup you can scale Librenms for what you need, and even use the rest of the resources for whatever you please(because you have way more than enough resources)