mapleFU commented on PR #3027: URL: https://github.com/apache/kvrocks/pull/3027#issuecomment-2972260623
The idea general LGTM, but I also wonder, would the performance more depending on block_cache's `high_pri_pool_ratio` and `low_pri_pool_ratio` in `LRUCacheOptions`, since the large blob might easily effect the global block cache? For Blobs, previous mode would cache the items and fetching the blobs, if the memory is enough for these blobs, not caching the blobs is a bad idea. But when memory is not enough, this mode might utililize SSD's random read Current mode making blob items into cache, when memory is enough this just enhance the performance, but in other scenerio this would making depending on cache policies Generally I think we can add a server side config for this( maybe default enable?) , and allowing config this when users see regression. -- 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]
