sarkaramrit1993 opened a new pull request, #29390:
URL: https://github.com/apache/flink/pull/29390

   ## What is the purpose of the change
   
   `CompressibleFSDataInputStream` only overrode `read()`. Bulk reads fell back 
to `InputStream#read(byte[], int, int)`, which loops over `read()`, so the 
wrapped stream got hit one byte at a time. Operator state restore (list, union 
and broadcast state, including source split state) reads through this class, so 
restoring large operator state was very slow. The JIRA has numbers: 120 MB took 
~44s before and ~136ms with bulk reads.
   
   ## Brief change log
   
     - Override `read(byte[], int, int)` in `CompressibleFSDataInputStream` to 
delegate to the (possibly compressing) wrapped stream.
   
   ## Verifying this change
   
   This change added tests and can be verified as follows:
   
     - Added 
`CompressibleFSDataInputStreamTest#testBulkReadIsForwardedToDelegate`, run with 
and without Snappy. It writes 64 KiB, reads it back with bulk reads, checks the 
bytes, and asserts the underlying stream saw no single-byte `read()` calls. 
Without the fix the uncompressed case makes 65,536 single-byte reads and fails.
     - `OperatorStateBackendTest` and `StateSnapshotCompressionTest` still pass.
   
   ## Does this pull request potentially affect one of the following parts:
   
     - Dependencies (does it add or upgrade a dependency): no
     - The public API, i.e., is any changed class annotated with 
`@Public(Evolving)`: no
     - The serializers: no
     - The runtime per-record code paths (performance sensitive): no
     - Anything that affects deployment or recovery: JobManager (and its 
components), Checkpointing, Kubernetes/Yarn, ZooKeeper: yes (faster operator 
state restore, no behavior change)
     - The S3 file system connector: no
   
   ## Documentation
   
     - Does this pull request introduce a new feature? no
     - If yes, how is the feature documented? not applicable
   


-- 
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