Notice: This project has migrated over to Unipop

elastic-gremlin

TinkerPop 3 implementation on Elasticsearch backend. You should read up on Tinkerpop before you use elastic-gremlin.

Features

• Scalable
  Using ElasticSearch's scale-out capabilities we can spread out our graph to many nodes, enabling more data while retaining good performance.
• Indexing
  We utilise ES's great indexing capabilities. Either let elastic-gremlin automatically create them, or configure the mappings for your specific needs.
  You can index Text (including analyzers), Numbers, Dates, Geo (just use the Geo predicate in a 'has' clause), etc..
• Custom Schema
  ES offers many different ways to customize the way your data is stored, enabling you to optimize it for your specific querying needs. We give you the power to use all these features and get the most out of your ES cluster.
  You can also utilize this ability to query existing data that you've loaded into ElasticSearch, by mapping the data to vertex-edge relationships of different kinds.
• Aggregations (Coming Soon)
  Aggregation traversals (e.g. g.V().count()) can benefit greatly from ES's Aggregation module

Getting Started!

1. clone & build elastic-gremlin

   git clone https://github.com/rmagen/elastic-gremlin.git

   mvn clean install -Dmaven.test.skip=true

2. Create an ElasticGraph:

   BaseConfiguration config = new BaseConfiguration();
   /* put configuration properties as you like*/
   ElasticGraph graph = new ElasticGraph(config);
   GraphTraversalSource g = graph.traversal();
   g.addV();

3. Or just use the Gremlin Server or Gremlin Console.

Confiuration

Basic

Basic usage of elastic-gremlin creates or uses an existing ES index, with each Vertex and Edge contained in its own document. You can customize some of the behaviour:

• elasticsearch.client (Default: "NODE")
  The client type used to connect to elasticsearch.
  • NODE Sets up a local elasticsearch node and runs against it. elastic-gremlin defaults to NODE, so you can get up and running as quickly as possible.
  • TRANSPORT_CLIENT Connects to an existing ES node.
  • NODE_CLIENT An optimized way to connect to an ES cluster. For more information read here
• elasticsearch.cluster.name(Default: "elasticsearch")
  The elasticsearch cluster's name.
• elasticsearch.cluster.address (Default: "127.0.0.1:9300")
  The elasticsearch nodes' address. The format is: "ip1:port1,ip2:port2,...".
• elasticsearch.refresh (Default: true)
  Whether to refresh the ES index before every search. Useful for testing.
• elasticsearch.index.name (Default: "graph")
  The name of the elasticsearch index.
• elasticsearch.bulk (Default: false)
  Cache all mutations in-memory and execute them in bulk when calling ElasticGraph.commit().

And most importantly you can customize the ES Index's Mappings to best fit your data. You can use ES's own APIs to do it. elastic-gremlin will automatically utilize your indices as best as he can.

Advanced

In addition to index mappings, ES offers many other ways to optimize your queries.

• Model your documents in different ways (Nested Objects, Parent-Child Relationship, etc)
• and your indices
• routing
• batch together queries
• upsert documents
• and any other ES feature that could help optimize your use-case...

Implement QueryHandler to use a customized schema that works best for your data.
We still don't have enough documentation on this, but you can take a look at the implementations of SimpleQueryHandler and ModernGraphQueryHandler

You're welcome to send us any comments or questions (rmagen@gmail.com)