I've got indexes storing the same kind of data split into weekly chunks - there has been some fairly substantial variation in data volume. I've got a mapping change I need to make across all the back data, and I'm thinking it might make sense to try to rebalance the documents per shard so that I have around 1 shard per N documents. Is that a worthwhile time investment in terms of query performance, or should I just stick with the 3 shards per index I've been using so far? I'd keep 3 shards as a minimum, so if there's a week with 10 documents it would still have 3 shards.
If I have an index that would end up with more than one shard per data node, does it make more sense to limit the number of shards to the number of data nodes, or go ahead and follow the 1 shard per N documents pattern? Thanks! Josh -- You received this message because you are subscribed to the Google Groups "elasticsearch" group. To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/d775acd2-9717-4fe6-bbda-9c6d42f0cb39%40googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
