I am new to Fink and Gelly. We have a requirement to do graph computation while the graph itself is constantly changing due to incremental updates (frequently). Because the graph is so huge, we did some investigation on Spark with TinkerPop. But the problem with this approach is that the RDD is immutable, which makes the incremental updates on the graph a kind of difficult to handle (IndexRDD has not been maintained for a long time so we do not want to go this direction). This makes us to think about Flink and Gelly. Can any experts help me answer these questions: 1. Gelly/Flink can handle incremental updates natively, right? What I mean is that graph traversal -> graph mutation -> graph traversal does not need extra efforts like spark to reload the entire RDD, right? But if so, how the data consistency is maintained in Flink? 2. What will happen if I update the graph (like add an edge, change the value of a vertex) while the graph computation is processing?
Thanks a lot
There is a github project for graph streaming