Performance benchmark

HI,
I am evaluating open source BPM engines. I need to run around 15000 workflows at any point of time.
Request you to point me to relevant documents if any work in this direction is done for Camunda.
Thanks

Hi,
This link may be of interest…

regards

Rob

Hi @Ashutosh_Kumar
this is a relatively generic question.
Performance depends on many different factors.

Some pointers:

  • How do your workflows look like? Are they mostly service orchestration?
  • How many save points / transaction boundaries do you have?
  • Which database do you want / have to use?
  • Do you need history of the workflows that were executed?
  • What do you exactly mean by “run around 15k workflows”? Is this being able to start 15k instances / second or is it that you expect that we can have 15k workflow running at the same time in the workflow engine?

Please give us more details and then we can see.
Generally speaking camunda bpm is one of the fastest workflow engine on the market.
This is also proven already by some research papers e.g. http://www.iaas.uni-stuttgart.de/RUS-data/INPROC-2016-05%20-%20Micro-Benchmarking%20BPMN%202.0%20WfMS%20with%20Workflow%20Patterns.pdf

Best
Felix

Hi,
Please see my responses below

  • How do your workflows look like? Are they mostly service orchestration?
    Yes mostly service orchestration with some tasks for human approvals , parsing of responses etc.

  • How many save points / transaction boundaries do you have?
    This is not yet decided but typically may be about 10

  • Which database do you want / have to use?
    I can use any open source database

  • Do you need history of the workflows that were executed?
    Yes history is needed

  • What do you exactly mean by “run around 15k workflows”? Is this being able to start 15k instances / second or is it that you expect that we can have 15k workflow running at the same time in the workflow engine?
    Starting can be staggered over a period of time . But all of these workflows may be running parallel.

Thanks

Ashutosh.