Details
Details
- Reviewers
cmollekopf - Commits
- R9:620c4f551d3d: Use Key API in indexes
Diff Detail
Diff Detail
- Repository
- R9 Sink
- Lint
Automatic diff as part of commit; lint not applicable. - Unit
Automatic diff as part of commit; unit tests not applicable.
Comment Actions
Some disk statistics:
Account 1: a lot of mails, but mostly text (occasional big attachments)
Account 2: a good number of mails, but mostly HTML heavy messages, with images, etc.
Comment Actions
Results of the Dummy resource write benchmark:
Some interesting bits:
Run benchmarks
Before | After |
---|---|
Current Rss usage [kb]: 40700 | Current Rss usage [kb]: 38484 |
Peak Rss usage [kb]: 40700 | Peak Rss usage [kb]: 38484 |
Rss growth [kb]: 15920 | Rss growth [kb]: 13660 |
Rss growth per entity [byte]: 3260 | Rss growth per entity [byte]: 2797 |
Rss without db [kb]: 29736 | Rss without db [kb]: 29504 |
Percentage peak rss error: 0 | Percentage peak rss error: 0 |
On disk [kb]: 10788 | On disk [kb]: 8804 |
Buffer size total [kb]: 898 | Buffer size total [kb]: 898 |
Write amplification: 12.0075 | Write amplification: 9.79923 |
Test Disk Usage:
Before | After |
---|---|
Free pages: 412 | Free pages: 312 |
Total pages: 760 | Total pages: 607 |
Used size: 1425408 | Used size: 1208320 |
Calculated key + value size: 856932 | Calculated key + value size: 702866 |
Calculated total db sizes: 970752 | Calculated total db sizes: 950272 |
Main store on disk: 3112960 | Main store on disk: 2486272 |
Total on disk: 3293184 | Total on disk: 2666496 |
Used size amplification: 1.66339 | Used size amplification: 1.71913 |
Write amplification: 3.63268 | Write amplification: 3.53733 |
Comment Actions
I'd say so too! But I'm still not sure why the disk statistics are so underwhelming :-/
Comment Actions
Yeah, those seem wrong to be honest. The only way those numbers make any sense if the uid size is completely dwarfed by the rest of the data, but that has not been our experience so far.