Tinkerpop3 connect to remote TitanDB server

Question

I am trying to obtain Graph object using Tinkerpop3 in Java as client from already running TitanDB server (I do not want to create server).

In other words, I am trying to implement such function:

public Graph obtainGraph(String serverIp, String graphName);

I was trying to do it like here: AWS Lambda + Tinkerpop/Gremlin + TitanDB on EC2 + AWS DynamoDB in cloud

but as I understand, TitanFactory.open() starts server, and I do not want to do this - I just want to connect to existing server.

Documentation as well as most materials in Internet use in-memory graphs for examples, and I cannot find one, that shows how to:

• create new graph and save it on remote server

• retrieve existing graph from remote server

• update such remote Graph, so after adding/removing edges committing changes

• delete whole graph

I do not want to do above stuff through Gremlin language (Strings), but through Java API (TinkerpopBlueprins). This guy is getting close to what I need: Add vertices to TitanDB Graph in Java however, his method already takes Graph as argument.

I have seen in many places in Internet, that GraphFactory.open() gets path to properties file, however I haven't seen example of content of such file, especially with TitanDB relevant data, so I would prefer to use Configuration object.

Graph graph = GraphFactory.open(new BaseConfiguration())

says, that there is no gremlin.graph property.

Configuration configuration = new BaseConfiguration();
configuration.setProperty("gremlin.graph", "titan");

Graph graph = GraphFactory.open(configuration);

says GraphFactory could not find [titan] - Ensure that the jar is in the classpath

Is there any statically typed builder with enums and constants, instead of Map<String, Object>, that will tell me, what properties I have to provide and what is their type? Is there any open source project, that uses Tinkerpop3 to connect as client to remote TitanDB server, that I could use as example?

I would like to see fully working example, instead of in-memory with external configuration.

Answer 1

Here is an example of a Titan driver program that connects to a running Titan Server. https://github.com/pluradj/titan-tp3-driver-example As you have noted, this will pass Gremlin as a string to the remote Titan Server.

If you don't want to do this because you want to use the Java API directly, you should use TitanFactory.open() to make a direct connection to your graph. TitanFactory.open() creates a TitanGraph instance that you can execute your graph API calls against. It does not start a Titan Server. Under the covers, it creates client connections to the backend storage and index.

You can refer to this example for a Titan Java program without Titan Server https://github.com/pluradj/titan-tp3-java-example

You can configure this using a properties file (here is an example configuration using Cassandra and Elasticsearch) or by constructing a Configuration object through code (basically setting the same key-value pairs that are in the properties file).

• If the graph does not exist before your initial connection, Titan will create the graph keyspace in Cassandra and index in Elasticsearch.

• Make note of the storage.hostname and index.search.hostname as these are your Cassandra and Elasticsearch clusters respectively. These are essentially your "graph server". You don't need a separate Titan Server running.

• Titan does not have any APIs to delete the graph from the backend storage. To delete the whole graph, you would need to connect to Cassandra through a Java client driver, and execute the API to drop the keyspace. Similarly, you would need to connect to Elasticsearch through its Indices API, and delete the index.