We are currently running hybrid network (on-prem and private cloud in Azure). Since CheckMK product is not HA-aware at the application level, it is only achieved at on-prem implementation by pairing Physical + Virtual Appliances (as per CheckMKofficial best practice). When it comes to the redundancy in the cloud, you must rely on the cloud-provided tools (e.g., Load Balancers, EKS as frontend, database replication in the backend, etc.). This brings complexity to the monitoring solution. I haven’t been able to locate the documentation at CheckMK site about that (deployed/supported architecture in cloud). As we run the hybrid network, on top of the workloads that are available in the enterprise edition, we also need the workloads that are also exclusively available in the cloud one as well.
My questions are:
Does CheckMK provide a cloud architecture scenario document on redundance?
Can CheckMK Cloud edition be installed on (Physical + Virtual) Appliances to leverage HA, assuming that on-prem + cloud agents are reachable over private network?
Will the checkMK Appliance accept installation of the cloud edition? Has someone ever managed to successfully implement it? If yes, would this deployed scenario be supported by CheckMK team?
This is a lot to unpack - considering that you are an Enterprise customer, it might be worth talking this through with whoever you bought Checkmk through (us or one of our partners).
The short answer to your question is:
I don’t think we have something right now, but it’s definitely something we should look into, because you won’t be the last one to ask for that.
Yes, I do think so.
I don’t see why this shouldn’t work. But maybe one of our consultants can confirm. I also believe that if you install any one of the commercial Checkmk Editions on one of our appliances, we would support that (assuming it is one of the versions still in support, not something ancient). But I’ll check with our support team on that, in case I’m unaware of something.
Regarding 2. and 3. (which sound the same to me): Yes, all commercial editions are supported on the Checkmk appliances, both virtual and physical. There are no downsides in running the CCE on the appliances.
What we have seen in the past is that the scale and use cases for which Checkmk is being used varies so much, that any general architecture or even just sizing recommendations more often than not is not right for the user / customer. It might be total overkill for many, and not advanced enough for others.
So we are refraining more and more from giving general architecture blueprints (of course it’s different when we work with individual customers as part of our consulting and support work).
Some general things we can say, of course:
If you are using the Checkmk appliances to cluster, only cluster two physical appliances (ideal) or a physical and a virtual appliance. Do not create HA clusters out to two virtual appliances (risking a split-brain scenario).
If you are running Checkmk on virtual machines, use your hypervisor’s features to establish the desired level of clustering / HA.
If you are running Checkmk in a public cloud environment, consider leveraging the redundancy features provided by the cloud provider to establish the desired level of clustering / HA. Given that choice of suitable solutions (and use cases, budget, …), we are not in a position to give specific recommendations.