As mentioned on the previous ZFS Leadership Call, Klara and iXsystems announced the start of work on the Fast Dedup project to address the performance of ZFS Dedup.
We're in the middle of a large coding effort and would appreciate other organizations joining as co-sponsors and collaborating with us. *Fast Dedup* will include a rearchitecting of the DDT to improve performance and efficiency. The new Fast Dedup Tables (FDT) have several significant new properties: * Dynamically sized to fit into RAM (Dedup Quota and automatic limiting for dedup vdevs) * Broken into smaller, more manageable chunks for efficient updating (Sharding) * Automatically prune non-dedupable data for better RAM efficiency (Dedup prune) * Massively reduce write amplification by using a log-based write process (Log Dedup) If you are interested in sponsorship and want to know more, please fill in this sponsorship information form <https://docs.google.com/forms/d/1dDj61SxfYeYPNOc943979fUNrUXpfT413EnoBUGO25U/viewform?edit_requested=true>. We'll then send you more detailed information and answer any questions. Thank you. ------------------------------------------ openzfs: openzfs-developer Permalink: https://openzfs.topicbox.com/groups/developer/T85351fcd05582c06-M14c6c62c0e312714630a039a Delivery options: https://openzfs.topicbox.com/groups/developer/subscription
