You have successfully applied the actor model to your software architecture. You have a robust and performant system, you are no longer busy synchronizing threads or writing try/catch/finally exception handlers. More of your time is now dedicated to better understand and model your domain, and actors are taking care of less creative tasks.

But your system is still running on a single server. Not only it will soon reach its hardware capacity, it is at constant availability risk. One of the reasons to choose the actor model was location transparency: frameworks like Akka let you seamlessly distribute the load between multiple cluster nodes. Is this what you need? When is it right time to support clustering? What does it take? And when can you manage without it?

We had to answer all these questions in our project at Norwegian Broadcasting Corporation (NRK) expanding our system to use Akka.NET cluster. Come to our talk and we share with you our experience and some of our secrets.