JunRuiLee opened a new issue, #687: URL: https://github.com/apache/paimon-rust/issues/687
### Problem The async Parquet reader currently uses a fixed 512 KiB metadata prefetch hint. This can read substantially more data than the footer metadata requires. For HDFS through OpenDAL's native service, positioned reads reuse a file-scoped reader. A local HDFS comparison showed better read performance when following parquet-hadoop's seek-based sequence instead: read the 8-byte footer, then fetch the exact footer metadata and requested page-index range. ### Proposed change - Remove the fixed 512 KiB metadata prefetch hint. - Keep page-index loading demand-driven. - Avoid loading OffsetIndex for an empty external row selection. - Add range-tracking tests for footer, metadata, and OffsetIndex reads. ### Validation Before the implementation PR is marked ready, document the HDFS comparison with: - the same dataset and query/read path for both variants; - file count and representative footer-metadata sizes; - repeated wall-clock results for fixed prefetch versus exact reads; - any available bytes-read or range-request counts. Implementation: #686 -- 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]
