[ 
https://issues.apache.org/jira/browse/OAK-10803?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17854399#comment-17854399
 ] 

Stefan Egli edited comment on OAK-10803 at 6/12/24 11:14 AM:
-------------------------------------------------------------

+1 to -both- all

1) I think in general we are more and more introducing new configs in different 
places but have different support on how that config can be modified. It might 
be useful to simplify and perhaps consolidate the different options..

2) +1. There was also the option to not even doing compression if it wasn't 
worth it - but the current code doesn't go that far at this point. Perhaps 
requiring a minimal memory saving config option would indeed be useful.

3) +1 for a metric


was (Author: egli):
+1 to both

1) I think in general we are more and more introducing new configs in different 
places but have different support on how that config can be modified. It might 
be useful to simplify and perhaps consolidate the different options..

2) +1. There was also the option to not even doing compression if it wasn't 
worth it - but the current code doesn't go that far at this point. Perhaps 
requiring a minimal memory saving config option would indeed be useful.

3) +1 for a metric

> Compress in-memory property values
> ----------------------------------
>
>                 Key: OAK-10803
>                 URL: https://issues.apache.org/jira/browse/OAK-10803
>             Project: Jackrabbit Oak
>          Issue Type: Improvement
>          Components: documentmk
>            Reporter: Daniel Iancu
>            Priority: Major
>             Fix For: 1.66.0
>
>
> Some properties in the repository can be quite large and consume a 
> considerable amount of memory. E.g. large multi-valued properties, blob IDs 
> that contain an inlined binary.
> With a certain size of the value, it may be beneficial to compress it when 
> loaded into memory/cache and uncompress it only on access. This shouldn't be 
> too difficult and can be implemented in a backward compatible way, because it 
> only affects the in-memory representation and not how data is stored.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to