build storage backend system specific mutations. now storage.MutationBuilder should be implemented for specific storage backend.
build fetch request and deserialize storage specific data into Edge/Vertex. now storage.QueryBuilder should be implemented for specific storage backend.
I think actual retry logic on hbase graph can be shared through multiple implementation when we add more backend system, but for now, I think above 2 was biggest point to seperate HBase specific codes from Edge/Vertex/Graph.
decouple following logics from graph client.
I think actual retry logic on hbase graph can be shared through multiple implementation when we add more backend system, but for now, I think above 2 was biggest point to seperate HBase specific codes from Edge/Vertex/Graph.
@daewon plz review this.