[
https://issues.apache.org/jira/browse/JAMES-2906?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16952659#comment-16952659
]
Benoit Tellier commented on JAMES-2906:
---------------------------------------
```
23.11 MSG indexed per second
4362 B/msg
Mean: 105ms
p99: 380
mean flags 150 ms
p99 flags 614 ms
```
VS
```
Reindexed/s: 22.46
Size per message: 5676 B
Mean: 108 ms
P99: 357 ms
mean flags 71 ms
p99 flags 158 ms
```
Impact of removing _source:
- 20M ElasticSearch disk space gain
- At the expense of flags update, significantly slower, and negatively
impacting other JMAP calls (getMailboxes p99 8x higher)
We believe it is not worth it.
> POC remove _source field in ElasticSearch
> -----------------------------------------
>
> Key: JAMES-2906
> URL: https://issues.apache.org/jira/browse/JAMES-2906
> Project: James Server
> Issue Type: Improvement
> Components: elasticsearch, mailbox
> Reporter: Benoit Tellier
> Priority: Major
>
> We can easily reduce indexation time & dataset size by a factor 2 by removing
> the _source field.
> Thus flags updates will then need to rely on full database read. (currently
> relying on patches)
> We need full statistics on the _source field removal impact in order to take,
> once and for all, a rational decision on this topic
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
