Sergey Shelukhin created HIVE-17424: ---------------------------------------
Summary: LLAP Parquet caching v2: IO elevator support Key: HIVE-17424 URL: https://issues.apache.org/jira/browse/HIVE-17424 Project: Hive Issue Type: Bug Reporter: Sergey Shelukhin It would be nice to move the reading, decompression etc. onto the IO threadpool. That would require some refactoring in Parquet codebase, or a considerable amount of duplicate logic. ORC path has a little bit of both - a lot is reused from ORC project but some logic (not code) to read the file is duplicated. Not sure how practical/acceptable to the community the refactoring might be in Parquet. -- This message was sent by Atlassian JIRA (v6.4.14#64029)