Is there a performance difference between querying 10 indexes with 1
shard, or 1 index with 10 shards?
How is the merging of the results be performed: Multithreaded in both
cases? Are there any low level differences when executing a search
between both cases, which could impact the speed?
Searching on 1 index with 10 shards or 10 indices with 1 shard has exactly
the same cost. At the end, a search is broken down and executed on each
shard it corresponds to (on different threads / nodes), and merged.
Is there a performance difference between querying 10 indexes with 1
shard, or 1 index with 10 shards?
How is the merging of the results be performed: Multithreaded in both
cases? Are there any low level differences when executing a search
between both cases, which could impact the speed?
On Fri, Jul 6, 2012 at 12:44 AM, Shay Banon kimchy@gmail.com wrote:
Searching on 1 index with 10 shards or 10 indices with 1 shard has exactly
the same cost. At the end, a search is broken down and executed on each
shard it corresponds to (on different threads / nodes), and merged.
Is there a performance difference between querying 10 indexes with 1
shard, or 1 index with 10 shards?
How is the merging of the results be performed: Multithreaded in both
cases? Are there any low level differences when executing a search
between both cases, which could impact the speed?
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.