Time Search with more node in one cluster

(Ngọc Phạm) #1

sorry, im newbie and i have one question.

i read in here https://www.elastic.co/guide/en/elasticsearch/guide/current/_scale_horizontally.html, it have :

  • “Increasing the number_of_replicas to 2”, the blogs index now has
    nine shards: three primaries and six replicas. This means that we can
    scale out to a total of nine nodes, again with one shard per node.
    This would allow us to triple search performance compared to our
    original three-node cluster.

i tried index database about 7526 hits,5 shards with index ,2 replicas, 3 node in one cluster , and SearchResponse result have 1313 hits.

I realize time search with one node or 3 node is same( response.getTook().getNanos() ) ,search in 1 node faster than 2 node, search in 2 node faster than 3 node.

My question is : Why? How to get max performance like Guide elasticsearch?

My PC run ubuntu 13.10,memory 8G, Intel® Core™ i3-4130 CPU @ 3.40GHz × 4 and use SSD

Thanks for your help.:smile:

(Adrien Grand) #2

I think your benchmark has two problems:

  • 7526 is so small that you're not really benchmarking search, just your network
  • the important metric that the page you linked is about is not latency, but throughput: the number of requests per second that your cluster can handle.

(system) #3