JanusGraph Picks Up Where TitanDB Left Off
Yesterday marked the formal launch of JanusGraph, a new Linux Foundation project formed to continue development of the TitanDB graph database.
JanusGraph is a fork TitanDB, the distributed graph database that was originally released in 2012 to enable users to find connections among large data sets composed of billions of vertices and edges. TitanDB was originally developed by a company called Aurelius, was acquired in 2015 by DataStax, the company behind the Apache Cassandra database.
Cassandra was one of the back-end database engines that TitanDB could sit on top of, along with Apache HBase and Oracle BerkeleyDB. Last June, DataStax unveiled DataStax Enterprise Graph, which constituted “a complete rewrite” of the database to tie TitanDB closer to Cassandra, DataStax executive vice president of engineering Martin Van Ryswyk told Datanami.
While DataStax incorporated the TitanDB graph capability into its commercial NoSQL package, the TitanDB community was growing unhappy with DataStax. In particular, the TitanDB community reacted strongly to DataStax actions that it perceived as blocking the continuation of TitanDB project at the Apache Software Foundation.
“After the DataStax lawyer blocked the move of the Titan project to the ASF, we took the project to the Linux Foundation under the name JanusGraph,” says P. Tayler Goetz, the PMC Chair of the Apache Storm project and a member of the technical staff at Hortonworks.
Misha Brukman, a Google Cloud Platform engineer and contributor for the new JanusGraph project, says TitanDB had stagnated since DataStax acquired Aurilius. “There have been no Titan releases since the 1.0 release in September 2015, and the repository has seen no updates since June 2016,” Brukman writes in a blog.
The idea behind JanusGraph is to “reinvigorate development of the distributed graph system to add new functionality, improve performance and scalability, and maintain a variety of storage backends,” he writes.
IBM Vice President of Open Technology Todd Moore welcomed the creation of JanusGraph in a blog post yesterday. “Forked from the latest TitanDB code, it will continue the legacy of Titan by providing a community-supported open source, scalable graph database with a variety of storage backends,” Moore writes.
JanusGraph incorporates support for the property graph model with the open source graph computing framework Apache TinkerPop and its Gremlin graph traversal language, Moore writes. “Over time, we anticipate that JanusGraph will become the de facto reference provider implementation for TinkerPop,” he says.
Last June, Big Blue introduced IBM Graph, a new cloud-based graph service based on Titan and TinkerPop that runs atop Cassandra and ElasticSearch. At the time, IBM executives said they wanted to avoid forking TitanDB. However, that sentiment appears to have changed.
According to the JanusGraph project’s homepage, the JanusGraph database supports Cassandra, HBase, and Oracle BerkeleyDB backend databases; ElasticSearch, Solr, and Lucene search engines; and integrates with Apache Hadoop, Apache Spark, and Apache Giraph.
Several JanusGraph project participants will be in Austin, Texas tomorrow for Graph Day Texas. The project is currently looking for coders to help develop and maintain the software.