About sharding data
If you are familiar with traditional relational databases, you are probably familiar with the concept of database sharding and may wonder if the same concept exists in Couchbase Server. There is a third internal, structure for organizing data in the Couchbase Sever; these structures are called vBuckets, an abbreviation for ‘virtual buckets.’ vBuckets are roughly functional equivalents of database shards for traditional relational databases. Unlike manual sharding which you may need to perform for relational database, the Couchbase SDKs automatically request updates on the location of vBucket information from Couchbase Server when you add nodes or perform failover.
vBuckets reference information across different records and distribute bucket information across a Couchbase cluster thereby supporting scalability, replicas and fail overs. Couchbase client SDKs abstract you from the level of vBuckets; your information storage and retrieval operations will be communicated between an SDK and memcached and Couchbase buckets.
In the background and at a lower level, Couchbase Server will automatically create, manage and update vBuckets; similarly your Couchbase SDK will also automatically request updates on vBucket information so that it can find information and store in the right place. In short, there is very little to worry about with vBuckets and most likely you will not be in direct contact with them. As a developer, you only need to be aware that vBuckets exist and the role they generally provide in the system.