Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 5 Feb 2010 20:51:15 +0100
From:      Lorenzo Perone <lopez.on.the.lists@yellowspace.net>
To:        Pawel Jakub Dawidek <pjd@FreeBSD.org>
Cc:        freebsd-fs@freebsd.org
Subject:   Re: Any news on the HAST Project?
Message-ID:  <25CA86F9-4B63-44FE-8A03-855DC1D4DF60@yellowspace.net>
In-Reply-To: <20100204210522.GB1733@garage.freebsd.pl>
References:  <914E8A1F-2FE9-4A7E-9BC7-6174402B57D3@yellowspace.net> <20100128132613.olxiwcq0go0g0w88@www.hmallett.co.uk> <9B1DF836-0CCC-4CB2-B83C-3040428A7344@yellowspace.net> <20100204210522.GB1733@garage.freebsd.pl>

next in thread | previous in thread | raw e-mail | index | archive | help

On 04.02.2010, at 22:05, Pawel Jakub Dawidek wrote:

> Every write request is handled synchronously by HAST. It is reported as
> completed only when it is stored on local and on remote node.

That's great for state - yet it also means that it won't be suitable
for some scenarios - such as secondaries connected over slow
links or with noticeably slower media. 

While I sort of imagine some of the tricky, messy implications 
of an aynchronous implementation, I was hoping for it somehow. 

That doesn't make HAST less magic - it just shifts its use 
to different scenarios than some I originally imagined, were
it was something more similar to what dfly's is doing with 
hammer mirror-stream - but at a filesystem-agnostic level
(wonder if that's even possible in theory)..

> This also makes ZFS great choice for running on HAST as 'zpool import'
> is very fast as oppose to fsck (at least until Jeff finish his SU+J
> project).

This sounds like great marriage. Could latency be further
enhanced by HASTing the pool geoms and using an 'unHASTed'
device as ZIL device, in theory (as soon as that's possible)?

Thanx a lot!

Regards,

Lorenzo




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?25CA86F9-4B63-44FE-8A03-855DC1D4DF60>