Large Partially-connected Erlang Clusters

Motiejus Jakštys
Backend Developer at Spil Games

Have you heard of a Node100 problem? Naïve Erlang clusters are complete network graphs which tend to blow up with a only a handful (100+) number of nodes. Solution is partially-connected mesh networks. The tricky part is easy communication for devs and transparent cluster management for devops.

In Spil Games n-tier architecture we use native Erlang message passing between tiers. Thus our cluster size (way over 500) requires partially connected network.

In this talk I will present a library which helps to create and maintain medium to large Erlang clusters. We have been using it for over 2 years now and time has come to announce and open-source it.

This talk is a part of the continuous effort Spil Games is making to give back to the community, as seen in Erlang Factory 2014 with the release of erl-cache and erl-memcache.

Target audience

Engineers interested or working with medium to large-scale Erlang clusters.


I believe elegant software makes our lives easier; I create elegant software to solve difficult problems. I am a scalable systems developer and a capable systems engineer. Currently I am busy with service-oriented architecture backend at Spil Games. 

Twitter: @mo_kelione
Github: Motiejus

Back to conference page