Elasticsearch parallel queries performance


We have been using elastic for a while ( running on version 1.7).
We noted that when a specific user is running a "heavy query" using the kebana all other users are experiencing a a huge performance delay it's like they are waiting the the "heavy" query to be complete.

so my question is :

  1. is this sounds reasonable to think that all the query run serial ?
  2. I plan to use anther type of data that's need to be more fast and reliable and should not be affected by other data queries.
    so if I will place it under a different index will it helps ?

Many thanks in advance.

No, queries run in parallel, but as they do compete for the same resources they can affect each other. A large dashboard with multiple aggregations can cause a lot of load on the system.

Placing it under a separate index may help as the shards may be smaller and might be cached more efficiently, but as long as they are on the same nodes as the other indices they can still be affected by the resource usage of heavy queries.


I will try to put the significant live high performance need data on a separate index.
is there a way to config the elasticsearch that a some index's will be in higher priority then others when racing on the same resource ?

I do not believe there is any way to prioritise certain types of queries or traffic over others in Elasticsearch.