Rajesh Balamohan created HIVE-23468:
---------------------------------------

             Summary: LLAP: Optimise OrcEncodedDataReader to avoid FS init to NN
                 Key: HIVE-23468
                 URL: https://issues.apache.org/jira/browse/HIVE-23468
             Project: Hive
          Issue Type: Improvement
            Reporter: Rajesh Balamohan


OrcEncodedDataReader materializes the supplier<fs> to check if it is a HDFS 
system or not. This causes unwanted call to NN even in cases when cache is 
completely warmed up.

[https://github.com/apache/hive/blob/master/llap-server/src/java/org/apache/hadoop/hive/llap/io/encoded/OrcEncodedDataReader.java#L540]

[https://github.com/apache/hive/blob/9f40d7cc1d889aa3079f3f494cf810fabe326e44/ql/src/java/org/apache/hadoop/hive/ql/io/HdfsUtils.java#L107]

Workaround is to set "hive.llap.io.use.fileid.path=false" to avoid this case.

IO elevator could get 100% cache hit from FileSystem impl in warmed up scenario.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to