Date: Fri, 27 Mar 2015 14:12:34 +1030 From: "O'Connor, Daniel" <darius@dons.net.au> To: Xin Li <delphij@delphij.net> Cc: "freebsd-hackers@freebsd.org" <freebsd-hackers@freebsd.org>, Adrian Chadd <adrian@freebsd.org>, Xin LI <d@delphij.net>, Wojciech Puchar <wojtek@puchar.net> Subject: Re: Seagate Archive HDD Message-ID: <FE0B3634-61C1-4A75-AB00-110B071A239F@dons.net.au> In-Reply-To: <5514D0A9.1040106@delphij.net> References: <alpine.BSF.2.20.1503261124380.1417@laptop.wojtek.intra> <55148E42.80708@delphij.net> <4F65B315-5FFE-4184-91FD-C05A40E0A26E@dons.net.au> <5514A133.8060409@delphij.net> <88DE4F68-B05E-4E4D-8C4A-DED8147172E7@dons.net.au> <CAJ-VmokEQjDXSHTn7o%2BEDnyrjDxckQkvWgQokKhhY1T1m3t_wA@mail.gmail.com> <5514D0A9.1040106@delphij.net>
next in thread | previous in thread | raw e-mail | index | archive | help
> On 27 Mar 2015, at 14:08, Xin Li <delphij@delphij.net> wrote: > On 3/26/15 17:58, Adrian Chadd wrote: >> ... all the talk about this stuff, and yet noone's talked about >> how ZFS is supposed to be append only, and boy wouldn't that be >> awesome. :) > > It doesn't overwrite existing data with the assumption that a, let's > say, 4K-native drive doesn't have a 1MB "band" when writing. > > I hope the "journal" would be helpful for this assuming it is on > non-volatile medium... The journal for the existing drives appears to be on the actual platter (but not shingled obviously). This gives from pretty suboptimal write latency as you can probably imagine.. -- Daniel O'Connor "The nice thing about standards is that there are so many of them to choose from." -- Andrew Tanenbaum GPG Fingerprint - 5596 B766 97C0 0E94 4347 295E E593 DC20 7B3F CE8C
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?FE0B3634-61C1-4A75-AB00-110B071A239F>