nickva commented on PR #5890:
URL: https://github.com/apache/couchdb/pull/5890#issuecomment-3943000500

   At @janl's suggestion ran an bulk_docs with and without this PR
   
   
   Used a k6 script 
https://gist.github.com/nickva/88564cf77763b2b000bd9563406f03dc
   
   ```BENCH_Q=16 BENCH_SCENARIOS=bulk_docs BENCH_BULK_DOCS_RATE=50 
BENCH_DURATION=3m k6 run k6_couchdb_constant_arrival.js```
   
   On an 8 core i5-8365U intel laptop with Ubuntu and Erlang 27.3.4.7. Ran the 
benchmark 3 times with before the PR main and after the PR.
   
   Main before the PR:
   
   ```
         { name:bulk_docs }...........: avg=29.96ms min=9.21ms  med=19.89ms 
max=294.76ms p(90)=50.89ms p(95)=84.26ms
         { name:bulk_docs }...........: avg=25.9ms  min=9.72ms  med=22.48ms 
max=122.82ms p(90)=39.45ms p(95)=49.31ms
         { name:bulk_docs }...........: avg=24.52ms min=9.9ms   med=21.36ms 
max=116.94ms p(90)=37.19ms p(95)=45.9ms    
   ```
   
   Main with the PR
   ```
         { name:bulk_docs }...........: avg=21.54ms min=9.44ms  med=19.12ms 
max=133.65ms p(90)=30.82ms p(95)=38.8ms
         { name:bulk_docs }...........: avg=21.96ms min=9.72ms  med=19.52ms 
max=128.56ms p(90)=31.65ms p(95)=39.28ms
         { name:bulk_docs }...........: avg=22.54ms min=9.32ms  med=19.97ms 
max=93.85ms  p(90)=32.67ms p(95)=40.78ms
   ```
   
   Average, p90 and p95 looks a bit better but overall it's not a large 
difference. It's probably dominated by disk IO times.


-- 
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