snvijaya commented on a change in pull request #2368:
URL: https://github.com/apache/hadoop/pull/2368#discussion_r503691466
##########
File path:
hadoop-tools/hadoop-azure/src/main/java/org/apache/hadoop/fs/azurebfs/services/AbfsInputStream.java
##########
@@ -178,11 +195,15 @@ private int readOneBlock(final byte[] b, final int off,
final int len) throws IO
buffer = new byte[bufferSize];
}
- // Enable readAhead when reading sequentially
- if (-1 == fCursorAfterLastRead || fCursorAfterLastRead == fCursor ||
b.length >= bufferSize) {
+ if (alwaysReadBufferSize) {
bytesRead = readInternal(fCursor, buffer, 0, bufferSize, false);
Review comment:
AlwaysReadBufferSize helped the IO pattern to match the Gen1 run. But to
be performant readAhead had to be enabled. For the customer scenario explained
in the JIRA , for the small row groups for an overall small parquet file size,
reading whole buffer size along with readAhead bought good performance.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
