I am testing deduplication in another server with 2VM-s windows server 2008 R2 and its not going bad writes and reads are fast enough as 2x2TB mirrored drives offer and seeing that same files are using 100% less space is very good to read it. I will keep filling up disks and see how it will go in next days.
ZFS Deduplication
- Thread starter bitblue.lab
- Start date
Hi bitblue.lab,
I'm very interested in your results. I saw a big impact when using approx. 300 GB of space after deduplication (>30 VMs). Random hangs up to stonithed nodes, even with 0.6.5.3. For small volumes it works great. I use it for backup combined with CoW images (once created, cloned to other machines and then used. Space for one and diffs for many using no deduplication at all). I do not know if the async delete problem is solved in current ZOL, but it was at least a big problem two month ago. I deleted a big image and the machine completely hanged and only resumed to work after the file was deleted (> 30min).
But please be aware, once enabled in your pool, you have to destroy your pool to get rid of everything deduplication related which could fill up your precious arc unnecessarily.
I'm very interested in your results. I saw a big impact when using approx. 300 GB of space after deduplication (>30 VMs). Random hangs up to stonithed nodes, even with 0.6.5.3. For small volumes it works great. I use it for backup combined with CoW images (once created, cloned to other machines and then used. Space for one and diffs for many using no deduplication at all). I do not know if the async delete problem is solved in current ZOL, but it was at least a big problem two month ago. I deleted a big image and the machine completely hanged and only resumed to work after the file was deleted (> 30min).
But please be aware, once enabled in your pool, you have to destroy your pool to get rid of everything deduplication related which could fill up your precious arc unnecessarily.