DevOps Stack Exchange is a question and answer site for software engineers working on automated testing, continuous delivery, service integration and monitoring, and building SDLC infrastructure. It only takes a minute to sign up.
Sign up to join this community
I'm running my application on 4 Virtual Machines with 8 3.2 GHz Virtual CPUs and 32 GB, however, I will be splitting up the processes into separate containers.
I am not sure how many containers to run per host. What are the typical numbers? For example how the VM to bare metal server ratio is 1:10 generally, any links on attributes to consider, decision frameworks to consider or experiences would help!
The same as any Linux load, consider no more processes in wait state than the number of CPU.
The load on 1, 5 and 15 mins given by uptime should ideally be 1 less than number of cores.
Containers are roughly isolated processes, leaving a core for orchestration avoid congestion.
That doesn't mean only 7 containers on a 8 cores machine, it's a matter of simultaneous load, so the best way is to measure it and adapt consequently.
I personally have spinned 100 containers on a 32vcpu machine and as many docker containers pin to random CPUs, there is no such correlation between number of containers and number of vcpus. As docker containers can be thought of as processes, it seems logical what @Tensibai has mentioned.
