HAST + ZFS + NFS + CARP

[Ben RUBSON]

> On 21 Jul 2016, at 09:51, InterNetX - Juergen Gotteswinter <jg at internetx.com> wrote:
> 
> i whould not expect to see something like maybe 2040 or so.
> 
> no offense agains the zfs devs, imho the fs itself is not the right
> place for this. I could be wrong, if yes someone please feel free to
> point at it.
> 
> In the End, to me it looks like you have taken the most important
> features out of hast, carp and probably rsf-1 cluster and with some
> mixing and stirring one can get a solution like this.

HAST (same for Linux DRBD) adds an additional stack between the disks and ZFS.
In addition, HAST may require a lot of network bandwidth depending on the pool layout, much more than the incoming data throughput.
Built-in ZFS replication would require not much network bandwidth than the incoming data throughput itself.
Suitable for long-distance replication :)

> no wont work, (somehow yes, but rather no: i dont know of anyone missing
> active sync replication, and manpower for this important parts is
> limited, but afaik theres the possility to sponsor such an addon.
> 
> like, shut up and take my money

We may then hope to see it before 2040 ;)

> Am 21.07.2016 um 07:52 schrieb Ben RUBSON:
>> 
>>> On 01 Jul 2016, at 17:02, Ben RUBSON <ben.rubson at gmail.com> wrote:
>>> 
>>> I think what we miss is some kind of this :
>>> http://milek.blogspot.fr/2007/03/zfs-online-replication.html
>>> http://www.compnect.net/?p=16461
>>> 
>>> Online replication built in ZFS would be awesome.
>> 
>> Note that I opened the following feature request a few days ago :
>> https://www.illumos.org/issues/7166
>> 
>> Could be interesting to follow it.
>> 
>> Ben