ajantha-bhat commented on PR #7105: URL: https://github.com/apache/iceberg/pull/7105#issuecomment-1482268557
> This makes me feel if we should really use a storage for this. Technically it sound like we should use a key-value store for (partition, spec-id) -> (statistics values) instead of a file-based storage. Maybe we should build an interface for it and just make storage a specific implementation, instead of saying it has to be backed by a file or multiple files, because it is has to be inefficient in read or write in that way. Just a thought. During the design phase, I did evaluate storing it in a KV store (like rocksdb or hbase index). It is also captured in the design document. But we concluded that we don't want to bring external dependencies into Iceberg core functions and also we cannot use the current clean-up logic with this. Hence, the File is chosen. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
