You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
we made some performance tests on search requests after our migration to the ES search engine (#742).. We still have some pretty slow response times (e.g over 5s on ~600k rows).
I looked into code and I see two main reasons of this latency :
Search results are stored into the database and the system trues to fetch all results (in an extra thread) from the DB even if only one result is required
Pagination is done at the "controller" layer instead of using the DB/Elasticsearch pagination capability. Because of that, a lot of data is fetched from the DB.
Primary tests I've done, show an improvement on response time and response time stability if search result storage is removed and pagination is used at DB layer.
I plan to make a PR with these modifications :
Move the Pagination from the controller to DAO layer
Disable the search result storage
This PR would at this time break a feature like *BaseSubscriptionInterceptor but we welcome more insight into how the feature is used currently to adapt our PR. This PR would also impact several method and interface signatures which will be updated.
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.
Hi !
we made some performance tests on search requests after our migration to the ES search engine (#742).. We still have some pretty slow response times (e.g over 5s on ~600k rows).
I looked into code and I see two main reasons of this latency :
Primary tests I've done, show an improvement on response time and response time stability if search result storage is removed and pagination is used at DB layer.
I plan to make a PR with these modifications :
This PR would at this time break a feature like *BaseSubscriptionInterceptor but we welcome more insight into how the feature is used currently to adapt our PR. This PR would also impact several method and interface signatures which will be updated.
We welcome any feedback !
this issue looks related to
#481
#536
The text was updated successfully, but these errors were encountered: