Hi i would like to create a cluster where nodes on different jvms(docker images) and on different machines and networks enter the cluster. Is that possible to implement with akka?
That is fine as long as all the nodes of a cluster can reach each other across the network.
Note that remoting/cluster is not meant as a general purpose network protocol, for example to do communication between services in a micro service architecture, but more of a way to communicate inside one micro service.
Using cluster for cross-service communication can easily introduce very tight coupling between unrelated services so that they must be upgraded lock step and all have the same version of libraries to be able to communicate etc.
Because of that, communication between different services is better done with a more explicit protocol like Akka HTTP or Akka gRPC.
Thank you I have designed the system with exactly that thought in mind.
But I have one more question if you could please kindly answer, regarding passing the data(Serializable objects able to pass in a message,usually small but not always) accross the system. i read from the Documentation about StreamRefs and you have that as API may change. Are there any other current alternatives fo such a case?(I know about changing the message size and about large message destinations, I just want to make sure i did not miss anything from the documentation)
StreamRefs are marked as may change, but the API has not changed for a long while and we have no current plans to change them.
For large messages, up to some point (2Mb from default config), the large message destinations is often enough, but for much larger objects it makes sense to reach for stream refs.
Reliable delivery also has a chunking feature that may be worth checking out, it splits up large serialised objects in smaller pieces and passed to a remote node before being reassembling them on the destination side: Reliable delivery • Akka Documentation