Grafana Alloy Agent - Capacity Planning

Grafana Alloy
1

Hello Team,

I’m planning to use Grafana Alloy Agent to monitor multiple DB’s. Any insights on capacity planning , regarding CPU, memory and storage that needs to be allocated based on # of DB’s instances monitored ?

2

How many metrics are you anticipating in total?

I don’t think there is a recommendation or calculation, but I’d say, unless you have hundreds of thousands of metrics, start small. Maybe 100MB or 200MB of memory and 0.5 CPU, and increase from there as required.

1 Like
3

Hello Tony,

Thank you much for the quick response.

We’re planning to use ( Oracle exporter ( 20-30 critical metrics) , mongo exporters ( 20 -30 metrics) , SQL Exporter ( 10 - 20 Critical metrics) on a single Grafana Alloy agent and multiple DB’s configured on same Agent to pull metrics ( 30 -50 databases collecting these many metrics ) .

Hence trying to understand how to plan Capacity in a similar case. What all should be taken in to consideration to plan this architecture.

  1. Is the resource allocation depends on metrics collected, processed and pushed? what should be a recommended and what would be the math to plan ( # of metrics and # of DB’s? )

  2. Is there any limit Alloy has to connect to multiple databases ?

4

There is probably a physical limit somewhere, but I don’t know what that is. You should reasonably expect one alloy node to handle at least a couple dozens of databases.

1 Like
5

Thank you !