Date: Thu, 12 May 2011 12:03:17 +0200 From: Alexander Leidinger <Alexander@Leidinger.net> To: =?utf-8?b?wqlpbXVu?= Mikecin <numisemis@gmail.com> Cc: freebsd-fs <freebsd-fs@freebsd.org> Subject: Re: ZFS: How to enable cache and logs. Message-ID: <20110512120317.12543g51m4im15k4@webmail.leidinger.net> In-Reply-To: <A0F4CB3E-1068-4E19-A50F-AA5620A9EB85@gmail.com> References: <4DCA5620.1030203@dannysplace.net> <4DCB455C.4020805@dannysplace.net> <4DCB81B8.6070301@digsys.bg> <A0F4CB3E-1068-4E19-A50F-AA5620A9EB85@gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Quoting =C2=A9imun Mikecin <numisemis@gmail.com> (from Thu, 12 May 2011 =20 10:02:42 +0200): > > On 12. svi. 2011., at 08:44, Daniel Kalchev wrote: > >> On 12.05.11 05:26, Danny Carroll wrote: >>> >>> - Don't use SSD for the Log device. Write speed tends to be a problem= . >> It all depends on your usage. You need to experiment, unfortunately. > > What is the alternative for log devices if you are not using SSD? > Rotating hard drives? > > AFAIK, two factors define the speed of log device: write transfer =20 > rate and write latency. There is also bus contention (either on the SCSI bus, or in the SATA =20 channel/controller, or on the PCI-whatever (e/X/y) bus). > You will not find a rotating hard drive that has a write latency =20 > anything near the write latency of even a slowest SSD you can find =20 > on the market. > On the other hand, only a very few rotating hard drives have a write =20 > transfer rate that can be compared to SSD's. And if your PCI-something bus is not saturated but your SCSI/SATA =20 controller struggles with the work which is thrown at it, a separate =20 log device (normal HD) on another controller could free up the =20 pool-controller(s) up to a situation where it can handle all requests =20 at full speed and the log-controller can provide the additional =20 throughput at full speed which the pool-controller was not able to =20 satisfy. What you do in this case is that you add more spindles (disks) =20 dedicated to sync-write operations. The normal RAID-common-knowledge =20 of "adding more spindles for more performance" applies here, just that =20 it is specially for sync-write operations. The generic hint to have =20 them faster than the pool-disks is an answer for the worst case. As =20 always, the worst case for one person may not be the worst case for =20 another persons workload. Bye, Alexander. --=20 If love is the answer, could you rephrase the question? =09=09-- Lily Tomlin http://www.Leidinger.net Alexander @ Leidinger.net: PGP ID =3D B0063FE7 http://www.FreeBSD.org netchild @ FreeBSD.org : PGP ID =3D 72077137
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20110512120317.12543g51m4im15k4>