Availability for production

Hello guys. Just wanted to ask to you, how reliable is the current state of Zeebe, for production.
I know it has been labeled as production ready, for a bit of time, now. I know this doesn’t mean
all the features are finished, which is normal.

I checked a bit some of the issues reported on Github and what I would like to know, if it is reliable from a Data point of view.

It would be understandable lets say, if there are some low frequency issues, that would require to restart the broker / cluster etc, but what I would need is to avoid data loss basically, to reach a situation in which I would have to recreate the broker, and its data, thus losing data about in progress process instances, especially let’s say if I will have some long running processes.

So to sum up, basically data loss is what I am a bit worried about, when thinking in using this technology to production. So any experience based views you have to share or you have some knowledge about this ? Is Zeebe reliable ?

THANKS !

1 Like

I would really appreciate some input. I am considering in using Zeebe for production projects, and I would need to understand what should be my expectations.

1 Like

Basically once you use Zeebe, and rely on it for production processes, its data kind of becomes critical, so that’s why my reasoning.

1 Like

Hi @raduone.

That is a reasonable question.

Should you use Zeebe in production? Well, I would say that it depends. Zeebe can be used for some use cases. We’ve customers that using Zeebe or Camunda cloud in production. However, it can be that your use case is currently not well covered.

We are working constantly on improving the reliability, especially the stability of the cluster.

Regarding the reliability of the data, I think we’re in a good position now. In the last releases, we improved a lot. Also, we try to be backward compatible to avoid data loss when upgrading to a newer version.

However, long-running workflows need additional considerations over short-running once. Currently, it is not possible to migrate workflow instance to a newer version of a workflow. This can be important if you need to change your workflow because of a business requirement or a bug in the workflow execution. There are ways to build your workflows in a way to support this but it’s a bit more advanced.

Best regards,
Philipp

1 Like

@philipp.ossler sorry for the late reply. Thanks for the information.
Basically my current use cases, are in such a way that I want
to use Zeebe as an engine for the execution of mostly relatively
short workflows. Basically I mostly used to allow the development
of distributed transactions across services - SAGA.

I have a functioning prototype and successful proof of concepts.
It’s safe to say that the client side part seems to be oke for my use cases.

I wanted to make sure that I will not reach situations in which I lose
the states, while running the Zeebe cluster.

Regarding the long processes, yeah it’s good to know that you cannot
migrate the instances to a new process version, so that is good to know.

1 Like

@philipp.ossler For short-lived SAGA, it shouldn’t be an impact that I cannot move instances,
to new workflow versions.

For long-lived processes, if I reach this use case and need it, I will document myself
to possible strategies, before taking the leap.