[GitHub] [flink-benchmarks] fredia commented on pull request #23: [FLINK-23399][state] Add a benchmark for rescaling

[GitBox <gi...@apache.org>]

fredia commented on pull request #23:
URL: https://github.com/apache/flink-benchmarks/pull/23#issuecomment-1086559519


   hi @pnowojski,  I updated this PR according to the suggestion above, would you please take a review again?
   I move some code to [flink-state-backends](https://github.com/apache/flink/tree/master/flink-state-backends/flink-statebackend-rocksdb/src/test/java/org/apache/flink/contrib/streaming/state/benchmark)  in https://github.com/apache/flink/pull/19177, and split the previous RescalingBenchmark into `RocksdbStateBackendRescalingBenchmark` and `FsStateBackendRescalingBenchmark`.  
   
   The results of those benchmarks are:
   
   ```
   # Run complete. Total time: 00:18:53
   
   Benchmark                                  (rescaleType)  Mode  Cnt      Score      Error  Units
   RocksdbStateBackendRescalingBenchmarkExecutor.rescale     RESCALE_IN  avgt   30  19825.918 ± 1261.372  ms/op
   RocksdbStateBackendRescalingBenchmarkExecutor.rescale    RESCALE_OUT  avgt   30   1811.478 ±   82.262  ms/op
   ```
   
   ```
   # Run complete. Total time: 00:17:11
   
   Benchmark                                     (rescaleType)  Mode  Cnt      Score     Error  Units
   FsStateBackendRescalingBenchmarkExecutor.rescale     RESCALE_IN  avgt   30  15890.784 ± 461.543  ms/op
   FsStateBackendBenchmarkExecutor.rescale    RESCALE_OUT  avgt   30   7858.031 ± 276.433  ms/op
   ```
   
   In practice, most rescaling scenarios are recovering from a large state(>=1GB). So I set the state to be larger, the rule of “keep the single benchmark invocation around/under 1 second” is not satisfied,  but the total run time of each benchmark is under 30 minutes. 
   


-- 
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: issues-unsubscribe@flink.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org