is dedupe feature pool-wide or fs-wide?

[Freddie Cash]

On Thu, Sep 6, 2012 at 11:06 AM, Zeus Panchenko <zeus at ibs.dn.ua> wrote:
> please, forgive me this stupid question, but is dedup feature pool-wide
> or fs-wide indeed?
>
> what I mean is: if I have dedup=on on two fs, than while copying one
> file to the fist one and after that to the second one, I will or will
> not have data deduplicated on the second copying?
>
> am I correct to think if dedup is block-level then it is pool-wide?

It's pool-wide, but can be enabled/disabled on a per-dataset basis.

IOW, ZFS scans all dedup-enabled datasets for duplicate blocks.

So, copying from one dedup-enabled filesystem to another
dedupe-enabled filesystem will generated duplicate blocks that should
end up with only DDT updates and not use any extra storage space.

-- 
Freddie Cash
fjwcash at gmail.com