Hi guys, i was looking into how many clients can a fleetDM server handle, and i found this answer from @zwass from 2 years ago: "The server ought to be able to handle ~20k endpoints. You'll find the web UI really slows down after about 5k, which we know how to fix but have not prioritized as of yet. I think (but don't know) that it would really struggle with ~200k." Does this still stand as true?

No! We have made dramatic improvements in Fleet 3.x that allow scaling into the hundreds of thousands.

oh sweet! i was looking for that info somewhere, but could not find it

https://github.com/fleetdm/fleet/blob/master/CHANGELOG.md#kolide-fleet-300-jul-23-2020 is the changelog and https://github.com/fleetdm/fleet/blob/master/docs/infrastructure/faq.md#has-anyone-stress-tested-fleet has a tad bit more info.

Eventually 15k~ we are still using Kollide Fleet 3.0.0 but I see that performance improvement is already there. Anyway we woll be looking to migrate to FleetDM soon

Great! You should have no trouble at all at that scale. Let us know if you do run into anything.

Great, that’s our actual use case, but we may be looking at bigger deployments soon, I’ll let you guys know how if there is any interesting insight to share

🍻

@zwass Sorry for going back to this topic. Do you have any rough estimate on what is the number of endpoints that a single fleet server may support? I mean with no LB at all in the middle. I am sure there is a big question mark depending on the queries being excuted tho...