Dimensioning of a HiveMQ CE Broker

I am dimensioning a HiveMQ CE broker on Ubuntu 20 with following connections and peak traffic:

  • Total 500 PUB client ids,
  • Total 500 SUB client ids,
  • Total 2500 topics
  • 5MB per day per client message size
  • One message every 60 seconds per topic
  • Average size per message : 800 Bytes
  • 3 Million messages per day
  • max queue size: 300,000 messages (not sure how close this will be to reality)

I want to know how to arrive at ballpark numbers for RAM, vCPUs, Bandwidth and disk storage.
I use Digital Ocean as my cloud infra provider but the answer would be valid for AWS as well.

Appreciate if you can also mention what heuristic/thumb rule you used to arrive at the capacity.


Hi @sanjmeh,

giving a rule of thumb is always hard to give because the needed capacity always depends so much factors.

But in your case I think you are definitely good to go with the minimum requirement aka 4 CPU’s 8gb RAM (in AWS terms this would be a c5.xlarge ec2 instance).

Michael from the HiveMQ team