Im building an Icinga Infrastrucure from Scratch, and i wanted to know if someone has experience with seperating systems for example: 1 Server for each system: Icinga2 Core (1 Server), Icingaweb2 (1 Server), Icinga Satellites (2 Server), Grafana (1 Server), and 1 Server for the Databases.
My goal is to reduce the data traffic as much as possible but i dont know where it makes sense to put systems together and where not.
Im going with Director so i guess director has to be installed on the icingaweb2 server.
Surely if your goal is to reduce data traffic as much as possible, you should
install everything on one machine, so that all inter-process communications
happen locally and not over the network?
Or did I misunderstand what you meant by “reduce data traffic”?
Maybe the best way is to have eveything on one server - as @Pooh wrote. But what is about load balancing and HA?
The more nodes, the more communication takes place in the cluster. If you turn on the debug node on every node, you’ll see how much communication is between every icinga cluster node. Also icingaweb2 is calling the API from icinga core, if somebody is clicking arround / auto refresh the check results. This also means that the more colleagues have icingaweb2 open, the more the API is called.
We fore example have two master servers on every master is installed icingaweb2 and the local database. On the master 1 is also installed the director. So icingaweb2 on master2 is a failover.
If it would be necessary to have the DB also in HA it gets more complicated and there would be more traffic for db sync as well.
So maybe you’ll have to find a way of balance between reduce network traffic and having HA/load balancing, depending on the requirements for your company.