Partisan: Enabling Cloud-Scale Erlang Applications.

In this work, we present an alternative distribution layer for Erlang, namedPartisan. Partisan is a topology-agnostic distributed programming model anddistribution layer that supports several network topologies for differentapplication scenarios: full mesh, peer-to-peer, client-server, andpublish-subscribe. Partisan allows application developers to specify thenetwork topology at runtime, rather than encoding topology-specific concernsinto application code. Partisan additionally adds support for more channels,enabling users to distribute messages over multiple channels, sometimes inparallel.

We implement and evaluate Partisan in the Erlang programming language and useit in the evaluation of three scenarios. The first scenario compares the rawperformance between Distributed Erlang and Partisan, and shows that Partisanperforms on par with or better than Distributed Erlang. The second scenariodemonstrates that distributing traffic over multiple connections enablesPartisan to perform up to 18x better under normal conditions, and up to 30xbetter in situations with network congestion and high concurrency. The thirdscenario demonstrates, using existing applications, that configuring thetopology at runtime allows applications to perform up to 13.5x better or scaleto clusters of thousands of nodes over the general-purpose runtime distributionlayer.