天蓝色Cosmos DB的图形遍历性能

Question

Cosmos DB allows us to store graph data using the gremlin query language.

Are there intelligent algorithms optimizing how the graph is split up among many servers? If not I can imagine some queries being extremely slow do to network latency between the shards.

Answer 1

The documentation is still a bit lacking, but there are some performance considerations for DocumentDb itself. Namely, setting up a PartitionKey that is adequately granular will split your data across multiple partitions, giving you higher throughput. You can find more here: https://docs.microsoft.com/en-us/azure/documentdb/documentdb-partition-data