Recently, I am trying to build my own graph database and graph compute engine system to support my research. The purpose of the system is to provide a distributed, low-latency store with computational engine based on RDDs. I named it "Morpheus", saluting another graph engine from Microsoft named "Trinity", which I took it's design concepts from this article and implemented in my system.
The system has three parts: a distributed in-memory key-value store, graph represent layer providing graph w/r operation functionality based on the k-v store and distributed data processing framework.
Right now, this project is under heavy development and I think it will took pretty much time until all of the parts are ready for missions. You can track the progress by visiting my GitHub profile. I will release some articles to reveal the details on the implementations. I have so much ideas in this moment and I need to test to make the right decision but I will provide reports in these articles and explain why I adapt those solutions.