gripleaf commented on PR #329:
URL: https://github.com/apache/paimon-cpp/pull/329#issuecomment-5710501086

   > I’d like to better understand two things here:
   > 
   > * What are the actual benchmark results for this optimization, and how 
does it behave across different data sizes?
   > * It seems that many kinds of caching could potentially be handled by the 
filesystem layer. If caching the resource here mainly reduces JSON 
deserialization, I’m curious what proportion of the total time is actually 
spent on that part.
   
   here is  the benchmark results under our specific scenario(point lookup + 
hdd hdfs), it has abort 6% improvement.
   
   | metrics | PR323 | + PR329 | + PR304/pre-buffer |
   | --- | ---: | ---: | ---: |
   | QPS | 71.9205 | 76.1071 | 54.9883 |
   | Mean | 55.169 ms | 51.893 ms | 71.985 ms |
   | P50 | 51.447 ms | 42.065 ms | 68.598 ms |
   | P95 | 73.872 ms | 83.655 ms | 83.911 ms |
   | P99 | 112.001 ms | 103.840 ms | 119.431 ms |


-- 
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]

Reply via email to