Stop the search request once it finds first 10 hits

Elastic Stack Elasticsearch
1

I am doing some performance testing on Elasticsearch. So I am comparing Oracle db with Elasticsearch.
The Oracle query is able to stop it's search once it finds the first 10matches by using the code

select * from mtd_sec_data
WHERE current_sighting_status= 'ACT'
fetch first 10 rows only;

I want to do similar action in Elasticsearch too. Any help is appreciable.

2

See the terminate_after parameter.

3

i have tried this query

GET /mtd_sec_data/_search?pretty&terminate_after=10
{  "query":{
      "match" : {
         "CURRENT_SIGHTING_STATUS": "ACT"
      }
   }
}

Even after testing the script it is having a similar throughput as in without using terminate_after. Does this really stops searching once it finds the first 10hits or is it searching the entire index?

4

How many total docs does the non-constrained search match?

It's also worth noting that queries now have optimisations for short-cutting non-competitive matches

5

It has around 536 hits and the index contains 1000docs.

6

It has around 536 hits and the index contains 1000docs.

Probably not a great benchmark on that sort of volume. Most of the time spent will be in JSON parsing/rendering, returning those 10 docs source. The matching part which is searching for a single search term with only ~500 uses will be a small part of the overall cost.

7

can we observe a clear difference by benchmarking on 1million data and above?

8

can we observe a clear difference by benchmarking on 1million data and above?

Perhaps clearer.

Benchmarking is a notoriously tricky thing what with things like caches playing a part (caches built into elasticsearch and file system caches).

We have a whole benchmarking framework and related forum topic

9

Thank you for giving an idea. It really gives some useful thoughts :slightly_smiling_face: