Large scale graph computation with Apache Giraph
Apache Giraph is a an implementation of Google's Pregel system for working with large scale graphs, going beyond the Map-Reduce model, with an approach tailored for graph algorithms. Whereas graph algorithms must be shoehorned into the Map-Reduce model, within Giraph they are expressed naturally by modeling the behavior of vertices across synchronized steps. Important algorithms such as PageRank become trivially simple, with orders of magnitude improvement over traditional Map-Reduce approaches. Giraph was originally developed at Yahoo! Research and is now an Apache project with a diverse and growing community, including committers from LinkedIn, Facebook, Twitter and several universities.
Watch the video of Jakob Homan talk here.