sgup432 commented on issue #16792:
URL: https://github.com/apache/lucene/issues/16792#issuecomment-6047046610
I did some experimentation and built a sandbox prototype
(`ALPDocValuesFormat`, opt-in per field) and measured it against the default
doc values format.
**Storage: real on-disk size, 100k values, bits per value**
| Dataset | Default | ALP | Ratio |
|:---|---:|---:|---:|
| Prices, 2 decimal places (e.g. `437.29`) | 64.02 | 20.28 | **3.16×** |
| Latency in ms, 3 decimal places | 56.02 | 20.12 | **2.78×** |
| Latitude, 6 decimal places | 64.02 | 28.29 | **2.26×** |
| Temperatures, 1 decimal place, small range | 8.18 | 8.27 | 0.99× |
| Random bit patterns | 64.02 | 64.24 | 1.00× |
Each value was indexed the way `DoubleField` stores doc values
(`SortedNumericDocValuesField` over `doubleToSortableLong`). The ALP index
routes the field to ALPDocValuesFormat through getDocValuesFormatForField.
Sizes are the `.dvd`+`.dvm` vs `.alpd`+`.alpm` files on disk, including all
metadata. The data is synthetic and unsorted.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
