3

I have implemented publisher-subscriber method for interprocess communication using Redis and ZeroMQ. What parameters should I consider to compare the performance of Redis and ZeroMQ?

I have read the answer given here - zmq vs redis for pub-sub pattern - but I want to test performance on my own machine and write my own methods to compare their performance.

JoshMc
  • 10,239
  • 2
  • 19
  • 38
xennygrimmato
  • 2,646
  • 7
  • 25
  • 47

1 Answers1

2

Performance comparison involves this among others:

  1. Both systems should be working in the same conditions. This means not running on the same hardware stack and operations system, backend databases (if required) etc. It also means that both systems should have compatible settings: f.e. for both systems you have to setup default settings. Or for both systems you have to setup it's maximum recommended settings. Ideal case if each system tuning will be performed by this system's profy.
  2. Don't forget about testing methodology. This means you have to warm-up each system (to fill caches etc), make as many as possible trials for each system (this will lead to a more accurate median or average parameter values)
  3. Use not selfmade testing tools / frameworks if possible (LoadUI, TestComplete, HP LoadRunner, Apache JMeter etc).
  4. Prepare the test scenario which will include the same steps for both systems. Try to send the same (or very similar data sets).
lospejos
  • 1,976
  • 3
  • 19
  • 35