https://github.com/orientechnologies/orientdb/issues/4105#issuecomment-101072462
I'm bringing up this issue, even though I've solved the primary problems, because it raised other questions that I think need to be answered or otherwise documented, specifically: Lessons Learned / Further Questions 1. Lightweight edges are not statistically faster with OrientDB 2.0.X and can't be indexed -- don't use them unless you're only using at most a handful of edges for any given node. 2. Groovy 2.4.3 is either significantly faster, or as fast as Groovy 1.8.9, depending upon the use case. - *Are there any known reasons why we shouldn't use Groovy 2.4.X for our Gremlin queries instead of Groovy 1.8.9?* 3. OrientBaseGraph::createIndex() is both awkward to use and under-documented. - *How do you use createIndex() to create an index with a composite key?* - *There's almost no useful documentation of the Parameter class.* - *What's the difference in performance / implementation of UNIQUE and UNIQUE_HASH_INDEX indexes?* 4. Vertex::getEdges() checks for useful indexes, but doesn't make use of the edge index I defined. - *Why not?* - *What indexes would be used by getEdges()?* - *How would I define them?* 5. Am I correct in surmising that edge data (eg: edge @rid <https://github.com/rid>) is stored on the source and target node records, in addition to on the edge record? - How do we tune this? - Is this more helpful for Gremlin or SQL queries? - What are the practical limits to scalability, for both SQL and Gremlin? 6. *What types of indexes could be defined to improve the performance of Gremlin queries when working with (or through) super-nodes?* -- --- You received this message because you are subscribed to the Google Groups "OrientDB" group. To unsubscribe from this group and stop receiving emails from it, send an email to orient-database+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.