On 5/31/2016 9:53 AM, Siddhartha Singh Sandhu wrote: > I was speculating whether sharding is done on: 1. index terms with > each shard having the whole document space. 2. document space with > each shard have num(documents/no. of shards) of the documents divided > between them.
If the router for the collection is "implicit" then sharding is 100 percent manual. You decide which shard gets the document when you index it. There is no automatic shard routing. If the router is "compositeId" then the shard is determined by doing a hash on the value of the uniqueKey field, then looking up which shard handles that hash in the clusterstate. This choice can be influenced by using a composite ID value. If there are plenty of documents and you don't use composite IDs, the distribution between shards will be mostly equal. The following URL contains some information on composite ID routing: https://lucidworks.com/blog/2014/01/06/multi-level-composite-id-routing-solrcloud/ Thanks, Shawn
