Clone this wiki locally
Welcome to the Giraph, large-scale graph processing on your existing Hadoop infrastructure!
2011.09.05 - Giraph has moved to the Apache Incubator! We will no longer be maintaining this GitHub repository. Our new site is at http://incubator.apache.org/giraph/
2011.08.09 - Completed initial release of Giraph with backwards compatible Hadoop 0.20.x (no security patches from 0.20.203). Try it out with '-Dhadoop=non_secure' when building.
2011.08.03 - Finished a series of tech talks on Giraph at companies like LinkedIn, Facebook, Turn, etc. We will be focusing on three upcoming changes for Giraph.
Backwards compatibility for Hadoop 0.20 users without security (i.e. Facebook append branch, Amazon EC2 versions of Hadoop)
- Support for different graph distribution methods (i.e. hash and range based)
Moving the code into Apache Incubator (a successful vote has been completed)
Thanks to everyone who came to see the talk at the Hadoop Summit this year. Many apologizes for tiny font, it certainly looked bigger on practice sessions. Here's a link to the presentation so that you can view it.
Here is the quick start guide to building Giraph and running a page rank benchmark:
Here is an example of a full Giraph application (including vertex input and output formats, input data, etc.) for single source shortest paths:
We're working hard on adding documentation and performance results. Stay tuned! By the way, please feel free to open issues if you run into them. We're looking to get a lot of input and help from the community. Also, please sign up for the mailings lists below that interest you. Until documentation is complete, the mailing lists are probably the best way to get your questions answered.
Please sign up for the mailing lists at http://incubator.apache.org/giraph/mail-lists.html
Our deprecated user mailing list history is available at http://groups.google.com/group/giraph-users . Please use for historical purposes only as all future posts should be on the Apache Incubator mailing lists.