JingsongLi commented on PR #9831: URL: https://github.com/apache/paimon/pull/9831#issuecomment-5772270765
Please add a reproducible end-to-end benchmark for the cold random-read claim. The current tests verify decoding correctness and that selected payload ranges cover fewer bytes than the full video. The byte-count assertion measures unique payload coverage; it excludes index reads and does not count repeated transfers. There is no indexed/unindexed latency or request-count comparison. Use representative long videos in object storage, the same decoder, and identical random target frames with and without the persisted index. Start cold runs with fresh decoder/cache state and report: - End-to-end first-frame and random-batch latency (p50/p95). - Actual total bytes transferred and GET counts, including index, metadata, media, and retries. - Additional ingestion time and index size from generating the index. This would quantify both the read benefit and the cost shifted to ingestion. -- 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]
