[
https://issues.apache.org/jira/browse/OAK-9311?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Fabrizio Fortino reassigned OAK-9311:
-------------------------------------
Assignee: Fabrizio Fortino
> oak-search-elastic: use low-level client for critical operations
> ----------------------------------------------------------------
>
> Key: OAK-9311
> URL: https://issues.apache.org/jira/browse/OAK-9311
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Reporter: Fabrizio Fortino
> Assignee: Fabrizio Fortino
> Priority: Major
>
> We are currently using the Java High-Level Rest Client [1] to interact with
> Elastic. While this is quite handy to use it does not seem to be the best
> option in terms of performance [2].
> Using the low-level client would mean to write more code in OAK with the
> advantage of having better performances and flexibility. For example, we
> could instruct ES to return only the response we are interested in [3]. In
> our searches, we are usually interested in the path field and, when requested
> the facets fields. ES for each document returns the index name. Since we use
> pretty long index names, only removing this would decrease the payload size
> by 20/30%.
> To have good performance and nice code we could keep using the query builders
> provided by ES. The query builders output should then be used to perform the
> request to ES through low-level client. The response needs to be parsed in
> OAK.
> [1]
> https://www.elastic.co/guide/en/elasticsearch/client/java-rest/master/java-rest-high.html
> [2]
> https://discuss.elastic.co/t/performance-difference-between-java-high-level-rest-client-vs-low-level-rest-client/125775
> [3]
> https://www.elastic.co/guide/en/elasticsearch/reference/current/common-options.html#common-options-response-filtering
> Options
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
