can anyone share, how much the number of agents that Kaseya can handle? i'm wondering, can Kaseya manage 200K agents/endpoints?
very appreciate for any information
We're just under 20K and went through hell for the past 3 months optimizing the database server and we're still getting "VSA Overloaded" warnings almost daily. This is on a DB server that is, by MSSQL standards, a beast with more than enough RAM, CPU, and all SSD storage.
Kaseya development is also working on a method to handle the fact that we hit the 2 billion record limit in the scriptlog table every 2-3 months which means we must purge the data and lose those logs.
So .. if this is what the reality is for 10% of your goal, then I'd venture to say you're looking at having many more than a single VSA instance.
Thanks PedroPolakoff3 for your sharing
i won the deal for 50K and based on your experience, i think i need to split the VSA into two instances.
and the problem is, next year our client adding another 200K endpoint, i have no idea how to build the VSA server for 200K
I suggest that you get your Kaseya rep involved and they should get level 2, 3 support involved. It is in their best interest ($$$ for 50K agents) to ensure you are successful. Let me know if they are not responsive and I will ping our rep. We are running on 1 VM for our VSA and 1 for SQL in Azure with 7700+ agents. MSPBuilder (Glenn Barnas) has some good info for ensuring performance as well. Go to: www.mspbuilder.com/downloads and register to access the Tech Brief "Optimizing the VSA Disk Layout"
Today you will need multiple instances. We are working on further power from the VSA framework to handle many more agents on a single instance.
Do you have any timelines Oscar? It's been many years since it's first been mentioned, and I have logged many ticket over the years around scalability & performance.
I have seen some vague diagrams, it would be nice to have some idea of when we can expect to see something we can use. We are nudging 10,000 agents and expect a lot more.
thanks guys for sharing
yesterday i was dropped my project for 200K, i'm not confident
Oscar, can you elaborate on what the maximum should be with the current on-premise platform?
Asep Suandi , congrats on the deal! I would be extremely cautious on trying to overload Kaseya servers. Hardware is pretty cheap if you are buying it and hosting it yourself, so spread the load out. Yes, it will be more complicated to support, but not impossible. What will really suck is if agents aren't checking in, getting false alarms because the server is over-loaded.
Here is Kaseya's current minimums for 20K machines: help.kaseya.com/.../Reqs.
I would guess the project will be broken up in some logical fashion, and you could break the Kaseya instances to match. If down the road you find that Kaseya has made it possible to support more agents on a server set, then you could always re-point the agents and consolidate.
If I can help with anything, let me know.
You can see below the recommended configuration for up to 20k endpoints.
Anything above that on a single instance would be a dedicated engagement with our professional services team.
From what I understand your team met with our top level engineer and developers to scope your particular goals with VSA. Were you present throughout these engagements?
Yes, we already make a concal today. so we take the 50K project and didn't take the additional (200K). maybe after Kaseya architecture stable, let say 1 instance can handle 100k endpoints
could you imagine if we take this offer, how many servers do we need to handle 50k+200k and another problem is how we distribute policy and everything else when the servers was distributed.
sure, will do. i'll keep posted
Thanks... I see Kaseya does not recommend SQL Server running on a VM past 20K agents. Is/has anyone checking/checked into the option of using Microsoft's Azure SQL (serverless) for those of us already using Azure?
FYI - SQL Cluster is supported to help distribute the load.
yes, next year maybe Kaseya VSA can handle >50K per instance
but right now it's better to stay 20K-25K per instance