Shard data based on associated social relationship

Abstract

A method includes identifying a plurality of content types in a database of a social network, wherein members produce and request content of different content types from the database. Job types to be performed on each content type are identified, when requests for content are received from members of the social network. A sharding process for each content type is selected based on the job types to be performed on the content types. The database is sharded into first group of shards for content types having job types that are parallel processing efficient and into second group of shards for content types having job types that are linear processing efficient. The first and second group of shards define a multi-shard database. Requests for contents are serviced from the first group or the second group of shards from the multi-shard database depending on the job types associated with the requests.