Solve the problem
It mainly solves complex relational network data storage, supports high concurrency, low latency, and high availability read and write capabilities
architecture
detailed architecture
storage format
vertisces: vertices
edges: edge
in-edge: in-edge out-edge: out-edge data redundancy, in order to increase performance
micro shards: micro-sharding performance evaluation | MongoDB Chinese Community How much shard is a machine
underlying structure
Full update
Mainly through the file storage system + MapReduce to improve efficiency.
batch update
The business generates a tag table based on the odps table, tells geabase which data is updated and which data is deleted, then geabase generates multiple odps sub-tables, and then flows back to the corresponding geabase table
Live Update
Write data synchronously through the message queue
the case
Find the intersection of F and FOFOF
Looking from the inside out,
first
NAV :Start = AccId :Edge_Type = 'associated' Query personal account
Then
Query friends with a relationship score greater than 80
personalized recommendation