Date: Thu, 23 May 2013 08:09:27 -0700 From: Michael Sierchio <kudzu@tenebras.com> To: Warren Block <wblock@wonkity.com> Cc: saeedeh motlagh <saeedeh.motlagh@gmail.com>, freebsd-questions <freebsd-questions@freebsd.org> Subject: Re: file corruption solution (soft-update or ZFS) Message-ID: <CAHu1Y704ufB_ie00hZrXw84WE5tT-b6t0O1uDjy69AbD7y9g-g@mail.gmail.com> In-Reply-To: <alpine.BSF.2.00.1305230629560.98184@wonkity.com> References: <CAN%2BS=WC0ThjT6ox9YdUsAb6%2B%2Bf=%2BqfCLuct_tN=jsr3KYTezaw@mail.gmail.com> <alpine.BSF.2.00.1305230554360.98184@wonkity.com> <CAN%2BS=WDGdL5ZabOGp%2BrBshjmrd9so5Q=wMc5sfDGuEAZxLe3%2Bg@mail.gmail.com> <alpine.BSF.2.00.1305230629560.98184@wonkity.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, May 23, 2013 at 5:33 AM, Warren Block <wblock@wonkity.com> wrote: > .. > One thing mentioned earlier is that ZFS wants lots of memory. 4G-8G > minimum, some might say as much as the server will hold. > > Not necessarily so - deduplication places great demands on memory, but that can be satisfied with dedicated cache devices (on SSD for performance and safety reasons). Without dedup, the requirements are more modest. Softupdates guarantee metadata consistency, but do nothing to address data integrity. ZFS has copy-on-write semantics (which solve a problem that even hardware RAID can't), and end-to-end checksums to detect/prevent data corruption (large drives will have uncorrectable bit errors over their lifetime). - M
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAHu1Y704ufB_ie00hZrXw84WE5tT-b6t0O1uDjy69AbD7y9g-g>