There are a lot of great references and recipes in the TinkerPop docs. One thing that would be nice to have is a simple explanation of what it means to use a TinkerPop implementation and what that guarantees. People coming from a relational world may misunderstand what to expect and might think they can do the equivalent of plugging in a new JDBC driver and away they go.
Some things to include
- How traversals are implementation agnostic with some caveats
- data types
- Connection options - different implementations may allow you to use TinkerPop drivers but may have implementation specific drivers for things like security or cluster awareness
- How to migrate easily between TinkerPop implementations should the need arise