Aggregation on flattened fields slows down due bulk writing


While working with aggregation on flattened fields, we encountered a problem.
We have a cluster, with 3 nodes, number of shards - 10, number of replicas - 2 and ~7 mln documents.
Average aggregation (with flattened fields) takes 300-500ms, but when we enable kafka consumer which begins bulk writing - aggregations on flattened fields significantly slows down to 7-8 seconds.

So constant bulk writing affects on aggregation only on flattened fields, what could be causing this situation and how to fix it?


What is the specification of the cluster and hardware used? What type of storage are you using? What does iostat -x show if you query with and without indexing?

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.