#tinkerpop

After Marko Rodriguez’s Knowledge representation and reasoning with graph databases, another great intro to graph databases resource is Joshua Shinavier’s presentation:

During a recent vacation between jobs, FoundationDB beta community member Will Wilson jumped into a new project and used our Java API to write a layer that implements the Blueprints 2.3.0 API. The entire layer took him two days to complete, including testing.

Blueprints is an open source specification of a graph database interface. The Blueprints API forms the lowest level of the Tinkerpop technology stack, which includes the Pipes framework for dataflow processing, the Gremlin query language, and the Furnace graph algorithms package. As a result, Will's layer allows you to use FoundationDB as a storage layer for the entire Tinkerpop stack.

Will has posted the FoundationDB Blueprints Graph API code to github. The layer passes the entire Blueprints test suite and supports most of the optional Blueprints features, including transactions for the graph database. Will notes that our Java API was an excellent match for Blueprints, and our transactions made it easy to implement features such as indexing that can otherwise pose tricky synchronization problems. For his next project, Will is thinking about implementing a new storage backend for Titan using FoundationDB.

The Blueprints layer illustrates the power of FoundationDB layers. Our core database combines an ordered key-value store with multikey ACID transactions, allowing multiple data elements to be updated in a single transaction. By mapping higher-level data elements to key-value pairs, a developer can write small amounts of code to implement a rich data model.

GraphDB-Bench is a collection of benchmarks for analyzing the performance of various graph frameworks

Anyone reviewed it? Anyone tried it? Is it a solid NoSQL benchmark?