Increase flatten performance of SearchSource #150658
Conversation
kertal
added
Team:DataDiscovery
kertal
changed the title
|
Pinging @elastic/kibana-data-discovery (Team:DataDiscovery) |
|
|
||
| const uniqueFieldNames = new Set(); | ||
| const uniqueFields = []; | ||
| for (const field of bodyFields.concat(filteredDocvalueFields)) { |
There was a problem hiding this comment.
We could also change to forEach here instead of for to minimize event loop blocking.
There was a problem hiding this comment.
thx, you mean?
bodyFields.concat(filteredDocvalueFields).forEach right? wouldn't this also block the event loop?
There was a problem hiding this comment.
Right. Well, it will block in chunks but it might also allow some browser work in between. Or maybe I am just making this up 😀
There was a problem hiding this comment.
I think it would be the case if it was an async/await type of forEach, but here I think it's blocking the even loop until done (pls prove me wrong 😃 )
|
@elasticmachine merge upstream |
|
@elasticmachine merge upstream |
💚 Build Succeeded
Metrics [docs]Page load bundle
History
To update your PR or re-run it, just comment with: cc @kertal |
kibanamachine
added
v8.8.0
backport:skip
Summary
This PR improves the performance of flattening the requested fields of the search source data request. it replaces the use of lodash
uniqWithwith a different implementation reducing the amount of iteration, which improves performance a lot when there are many fields. So when I tested with a data view having about 10.000 fields the time used for this loop went from 130-150ms to 2.3-4.1 msHowever it's important to note that this it just improved when there are field filters in the data view set. Else the improved part of the code isn't hit.
Fixes #136854