Hi everyone 
,
I’m exploring best practices for building distributed Erlang systems and would love to hear about your real-world experiences. I have three main questions:
1. Supervision Tree Distribution Strategy
When building distributed Erlang applications, how do you typically approach distribution?
-
Do you distribute entire supervision trees by application?
-
Or do you distribute only specific processes within the supervision tree?
2. Distribution Libraries and Extensions
Do you use the native Erlang distribution model, or do you leverage external libraries/extensions such as:
-
Partisan
-
SD Erlang
-
Other solutions?
If you use external tools, what motivated that choice over the native approach?
3. Production Scale
-
What’s the maximum number of BEAM nodes you’ve successfully run in production?
-
What lessons learned or challenges did you face at that scale?
I’d really appreciate any insights you can share from your production experience. Thanks in advance!