Date: Sun, 16 Oct 2011 21:40:09 +0300 From: "Luchesar V. ILIEV" <luchesar.iliev@gmail.com> To: Daniel Kalchev <daniel@digsys.bg> Cc: freebsd-fs@freebsd.org Subject: Re: [ZFS] Using SSD with partitions Message-ID: <4E9B2509.3030701@gmail.com> In-Reply-To: <14628DFB-AA3E-4D2D-9D4F-723B6327B6C0@digsys.bg> References: <CACh33Fpz=uAp8h0Bjsi1Be=ob_94jXtN51mAHvGPkReY5MpTcg@mail.gmail.com> <4E9AE725.4040001@gmail.com> <169E82FD-3B61-4CAB-B067-D380D69CDED5@digsys.bg> <4E9B1C1E.7090804@gmail.com> <14628DFB-AA3E-4D2D-9D4F-723B6327B6C0@digsys.bg>
next in thread | previous in thread | raw e-mail | index | archive | help
On 16/10/2011 21:09, Daniel Kalchev wrote: > > On Oct 16, 2011, at 21:02 , Luchesar V. ILIEV wrote: > >> On 16/10/2011 19:17, Daniel Kalchev wrote: >>> Therefore, with ZFS v28, adding ZIL does not introduce any more >>> risk to your data. >> >> I might be wrong in my interpretation, but from what I remember, >> when the power goes down, an unprotected SSD is likely to lose >> _more_ data than simply its write buffers -- that's quite unlike a >> hard-drive. So much, in fact, that the whole ZIL might become >> corrupted (and that's potentially way more data than any device >> cache). > > The real risk with low-grade "unprotected" SSDs is that the SSD may > well become damaged, sometimes beyond repair. > > It is the same risk with SSDs or with magnetic drives. If the drive > lies to the OS that it has safely written data -- then data will be > lost. Thing is, we know what a cheap HDD is. Most SSDs however lie, > because otherwise they will offer very poor write performance. That's true, but my understanding is that the differences go further beyond that. To quote one paper: "Our data show that flash memory’s behavior under power failure is surprising in several ways. First, operations that come closer to completion do not necessarily exhibit fewer bit errors. Second, power failure not only results in failure of the operation in progress, it can also corrupt data already present in the flash device. Third, power failure can negatively impact the integrity of future data written to the device." http://cseweb.ucsd.edu/users/swanson/papers/DAC2011PowerCut.pdf However, that's probably getting too academic (also beyond my own qualifications), and I wouldn't like to hijack the thread. > ZIL is not about RAM. ZIL is for low latency synchronous writing. It > does not matter how much RAM do you have -- it will not help if you > have heavy synchronous writing (of small records). >From what I understand, Patrick is talking about a home system, which is not very likely to be heavy on the synchronous writes, unless, of course, he's using NFS or a database. On the other hand, most desktop applications would happily use the additional memory, so it benefits not just the storage subsystem. That's why I'm making the point about the RAM upgrade, but, apart from that, you're absolutely correct about ZIL and synchronous writes. Cheers, Luchesar
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4E9B2509.3030701>