After discussing with Yangke, Lianxin, the second stage would focus on the code refactoring to tune the system performance before we move the distributed configuration. Therefore, we focus on the following two aspects.

1. where is the limit of the system perf ? vs datasize (number of vertex, edges)
2. what is the configuration parameter for the rocksdb or our computation engine to win the best pef ?
3. how can we tune our system to get the better pef ? code refactoring or others?

More details feature like

• Provide the system performance monitor and pef report
• Provide the graph API via Python and Grelimin support.
• Refactor the code and improve the readability and system perf.
• Provide graph security and access control. Finish the design doc.
• Support fault tolerance and auto restart.