Date: Fri, 17 Feb 2012 19:09:21 +0000 From: Alexander Best <arundel@freebsd.org> To: Julian Elischer <julian@freebsd.org> Cc: freebsd-fs@freebsd.org Subject: Re: freebsd-swap on ssd Message-ID: <20120217190921.GA26568@freebsd.org> In-Reply-To: <4F3E9A14.3070605@freebsd.org> References: <20120217141607.GA63659@freebsd.org> <4F3E9A14.3070605@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri Feb 17 12, Julian Elischer wrote: > On 2/17/12 6:16 AM, Alexander Best wrote: > >hi there, > > > >putting a freebsd-ufs partition on an ssd isn't recommended, since the ufs > >structure (unlike zfs e.g.) requires certain data to be continuously > >written to > >a fixed location and thus will cause the ssd to quickly run out of > >write-cycles > >and die. > nonsense. > the SSD doesn't use the same flash for the same logical locatio each time! > it maps it to different locations each time. i simply repeated what kirk mckusick said in the SU+J introduction video. he said for exactly this reason ufs should not be used on an ssd, since stuff like inode entries live in a fixed location, whereas with zfs the ueberblock can live in 128 locations. also in case of SU+J, where the journal only takes up a very small part of the disk due to the fact that it's only tracking metadata changes and isn't doing logging (like gjournal), there's also the chance to run out of write-cycles. see: http://youtu.be/_NuhRkiInvA cheers. alex > > >but how about using a small ssd (approx. 10GB) as one entire freebsd-swap > >partition? will this make more sense, or are there certain structures > >within > >the freebsd-swap partition type, which also need to be continuously > >written to > >a fixed location? > small SSDs may have less wear resistance than big ones.. the cheap > ones may not even > use proper mapping.. > >another question i'd like to ask: are there also issues with read-cycles on > >ssds? because i was thinking about putting a freebsd-boot partition on an > >ssd > >drive and only mounting it ro. this should solve the write-cycle issue in > >theory. however i'm not sure, if stuff like the dirty bit or the ufs label > >will > >also remain untouched. so even though the partition will only be mounted > >ro, > >freebsd might still frequently write certain data to a fixed location on > >the > >ssd drive which hosts the freebsd-boot partition. if this is the case, is > >there > >a way of completely prohibiting any writes to a disk? will revoking any > >write > >permissions from the device entry under /dev guarantee this, or is using a > >any > >device 100% ro under freebsd impossible (unless it has a hardware switch to > >forbid writes)? > > yes there are small issues with read cycles but it is all hidden from > you by the drive. > >cheers. > >alex > >_______________________________________________ > >freebsd-fs@freebsd.org mailing list > >http://lists.freebsd.org/mailman/listinfo/freebsd-fs > >To unsubscribe, send any mail to "freebsd-fs-unsubscribe@freebsd.org" > >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20120217190921.GA26568>