Date: Sat, 17 Dec 2005 11:55:06 +0100 From: "Poul-Henning Kamp" <phk@phk.freebsd.dk> To: Mathieu Arnold <mat@mat.cc> Cc: Alexandre DELAY <alexandre.delay@free.fr>, freebsd-hardware@freebsd.org Subject: Re: HDD Message-ID: <23984.1134816906@critter.freebsd.dk> In-Reply-To: Your message of "Sat, 17 Dec 2005 11:49:35 %2B0100." <0D162CCAE42F819501B87257@cc-126-240.int.t-online.fr>
next in thread | previous in thread | raw e-mail | index | archive | help
In message <0D162CCAE42F819501B87257@cc-126-240.int.t-online.fr>, Mathieu Arnol d writes: >+-le 14/12/2005 17:45 +0100, Alexandre DELAY =E9crivait : >| Don't you think that flash drives are also a good solution? >|=20 >| I am sure that it will be the future replacement for hard drive disks. >| see http://www.memtech.com/ for example >|=20 >| With no buffer and 1ms access delay you minimise write failures. It is an >| interresting solution. > >Hum, flash has a *limited* ammount of possible write for each cell, for low >cost, it's between 10K-50K and for heavy duty, industrial grade, bla bla bla, >it's around 2M, so, hum, just imagine you have a solid state flash disk, and >your swap on it. One day, your box begins swapping hard, and some time later, >half of your drive can't be written to again... Sorry but that is not how flash devices work. There is a very important piece of technology called a "flash adaptation layer" which performs what is usually called "wear levelling" in an attempt to get the maximum amount of lifetime out of the entire device by using a dynamic logical to physical mapping and keeping track of bad cells etc etc. Yes, you will kill a flash by doing a lot of writes, but it will not develop individual bad sectors like a disk. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 phk@FreeBSD.ORG | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?23984.1134816906>