Date: Sun, 15 Jun 2008 18:57:46 +0100 From: RW <fbsd06@mlists.homeunix.com> To: freebsd-questions@freebsd.org Subject: Re: md devices mounted with async Message-ID: <20080615185746.60efea11@gumby.homeunix.com.> In-Reply-To: <4854FF77.80901@FreeBSD.org> References: <20080614224742.17316919@gumby.homeunix.com.> <48545212.4040006@FreeBSD.org> <20080615013158.7dd19cf0@gumby.homeunix.com.> <48546B92.5050906@FreeBSD.org> <20080615035555.0b5d4b1c@gumby.homeunix.com.> <4854FF77.80901@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 15 Jun 2008 13:39:35 +0200 Kris Kennaway <kris@FreeBSD.org> wrote: > So I was referring to vnode devices. I guess there was some > confusion because in your reply you mentioned swap, not vnode. Sorry, that was my fault. I meant to ask about about malloc and swap backed devices, and never spotted that I'd written vnode, or that vnode had been mentioned. > Anyway, to be clear: when a filesystem mounted on md is written to it > writes through to its' backing store according to the mount policy of > the filesystem (e.g. with sync mounts all writes are written through > synchronously, etc). That is why async mounts on top of the md are > most efficient (as with mounts on top of any device). But isn't the normal point of async to avoid having the cpu waiting for the disk with nothing to do. If there's nothing to block on, the cpu can be consistently utilized without async. I suppose with SMP it might help spread the work between cpus, but with a single cpu it seems like it would just end-up doing the same work in a different order.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20080615185746.60efea11>