Date: Thu, 14 Mar 2024 19:39:32 +0000 From: Frank Leonhardt <freebsd-doc@fjl.co.uk> To: freebsd-hardware@freebsd.org Subject: Re: WD Blue 510 SSD and strange write performance Message-ID: <4C9E3F3A-90B3-43B1-B01F-E157E1107E1E@fjl.co.uk> In-Reply-To: <CCAB653B-4DC6-4C69-AB68-CD258200D22F@gid.co.uk> References: <e5c2a99d-931e-48b4-9445-fc4ad05ccc70@sentex.net> <CCAB653B-4DC6-4C69-AB68-CD258200D22F@gid.co.uk>
index | next in thread | previous in thread | raw e-mail
Flash storage is complicated. I doubt there's a huge cache in them, as as it would be volatile it'd be a big no-no for synchronous writes. The OS could cache it, of course. And if you're using ZFS then all bets are off. ZFS guarantees (for POSIX) that a synchronous write goes to non volatile memory before the system call returns. However this does not mean it goes to the final location on the disk. ZFS has an intent log of stuff that needs to be sorted out sometime, but is safe for now. So with can write away, especially with highly compressible blocks, and it'll go fast enough until it gets to the point it needs to get the FS in order (you run out of ZIL). I'm speculating about the cause of the effect in your case, but I'm not surprised you're getting an effect. Look up Flash storage strategies and the workings of the ZIL for more information.home | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4C9E3F3A-90B3-43B1-B01F-E157E1107E1E>