How does Titan stores data in HBASE

Question

I would like to know how graphs are stored when Titan DB uses HBASE as its data storage.

Why I am asking: assume that I want to leverage BOTH the Titan interface (ie interact with my data as a graph) and also be able, from time to time, to execute queries (say via Hive) by-passing Titan (the rationale for this could be situations in which I need more SQL-ish types of queries).

Answer 1

This page describes how Titan stores data in the various backends:

https://github.com/thinkaurelius/titan/wiki/Titan-Data-Model

The data that Titan stores within this model is encoded and compressed for optimal efficiency when reading and writing data. As a result, trying to access it with external tools will not work so well. For queries over the entire graph (which is what I think you are referring to with this question) use Faunus to either:

1. directly query the graph to get your answer
2. or use it to extract the data from Titan, transforming it to a format that might then be consumable by other tools.