Hi,
Absolute beginner here, also trying to evaluate what is there in camunda for us. First thing pops up in my mind, is that we have really huge and complex workflows say about 100 transitions and may be about 50 state and decision points including lot of timers.
Hence, anyone has any example project running really complex workflows and how is it scaling for you.
Complexity of a workflow is relative i think.I wouldn’t consider 50 states particularly complex. I’m not sure what you consider a transition so hard to say if that would add to the complexity. Are the specific things that you’re interested in know with regards to complexity.
Hi @sherry-ummen,
There are multiple companies/ organizations out there with complex workflows. What do you mean with transition?
If you create a lot of jobs (timers and asynchronous continuation) you should consider the configuration of your job executor. We have here a best- practice about scaling and performance. I hope it gives you some help to understand the basics of performance when it comes to Camunda.
If you face a performance issue with your workflow it is best to share the BPMN model, your architecture and your configuration. That would then give insides and help to identify bottlenecks.
I hope that helps
Nele
Thank you for the quick responses, that also tells about the community 
Ok, transition probably is more of the state machine terminology, which we are using. But now looking into the workflow direction.
By complexity I mean, there can be 100 or more states/transitions/decision points. Some communicating with external service and communications via Kafka topics.
Now we have series of events coming to our system, each event is passed through a workflow, so in a given time there is a possibility that events can launch the same workflow parallelly and can be in a different state.
And these events can be in thousands/sec/min (for now)
Now wondering, how camunda will behave in these scenarios, in terms of memory/cpu requirements
We also need these workflows to be easy to develop by developers by hooking up to new services and also the business analyst could create those workflows with ease.
All these sounds very much what camunda does already, but I need some more real world examples and opinions on this subject.
thanks
Generally the number of a states that a process definition has won’t have much of an affect on performance. But i think as you alluded to the number of state transitions per second required by the engine is the most important indication of performance.
Kafka is often integrated with Camunda, triggering processes or sending messages back to kafka so this isn’t unusual.
The workflows themselves are pretty easy to build for anyone really, but you should really give it a try first, there’s no other way to know for sure
Absolutely, thats what I am going to do next. Try it out . Thanks for the help