Date: Mon, 10 Nov 2008 18:56:03 +0000 (UTC) From: Szabolcs Szakacsits <szaka@ntfs-3g.org> To: freebsd-questions@freebsd.org Subject: Re: Sluggish scheduling during a long disk copy Message-ID: <loom.20081110T184409-956@post.gmane.org> References: <20081109202149.GA7091@ourbrains.org> <991123400811091225t392bd3f3i531dbe348a13e5e4@mail.gmail.com> <20081109203241.GB8395@ourbrains.org> <28283d910811091235q70181b52nc4235aea61518cd@mail.gmail.com> <49174EAC.2070403@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Kris Kennaway <kris <at> FreeBSD.org> writes: > Sounds like it to me. ntfs-3g uses FUSE, which is a userland filesystem > framework. By design it will have poor I/O performance since every I/O > transfer will require multiple trips into and out of the kernel. Performance doesn't work like that for file systems because typically the dominant factors are the file system design and the quality of the implementation. Even the still unoptimized ntfs-3g driver can far outperform other kernel file systems in streaming read/write speed (maximum ever measured sustained write speed is 902 MByte/s) and IO ops on Linux. Regards, Szaka -- NTFS-3G: http://ntfs-3g.org
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?loom.20081110T184409-956>