Date: Tue, 8 Feb 2011 10:50:30 -0600 From: Adam Vande More <amvandemore@gmail.com> To: Daniel Braniss <danny@cs.huji.ac.il> Cc: freebsd-emulation@freebsd.org Subject: Re: VirtualBox native iscsi initiator performance is bad Message-ID: <AANLkTin6gZiSenKK0MMQm1gfyO_YsV=-b9aDEUR-wKHg@mail.gmail.com> In-Reply-To: <E1Pmjc7-000I2Z-Qe@kabab.cs.huji.ac.il> References: <AANLkTinokgSOSQ_%2B3Zzj5mMz3MYf3npzuKACFsUR7oDC@mail.gmail.com> <E1Pmjc7-000I2Z-Qe@kabab.cs.huji.ac.il>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Feb 8, 2011 at 3:11 AM, Daniel Braniss <danny@cs.huji.ac.il> wrote: > > Using the native Vbox iscsi initiator for very simple tests I'm getting > > slightly less than 20 MB/s write speed and VM takes a long time to boot > > because of DMA LBA errors are being reported on the iscsi/sata disk. > > > > If I mount the iscsi on the host, I get close to 50 MB/s and then if I > map > > that to a Virtualbox raw disk, and stick in a VM I get around 42 MB/s > write > > speed. I haven't done any tuning on iscsi target so perhaps that could > > improve things, but I'm concerned about the huge drop off in performance > and > > reliability issues with the Vbox native version. Does anyone have > insight > > into this? > > > > virtualbox-ose-3.2.12 > > istgt-20110103 > > > > These are both recent 8-STABLE versions of Virtualbox. > > have you measured the tcp/ip performance of the virtual box? last time I > checked > it was not that good, remeber that iscsi is TCP based. > Yes, I have. I should have mentioned the VM I'm testing with is a 9-CURRENT with the virtio patch. However if I understand it correctly, the native vbox initiator isn't in the VM itself, it's at the hypervisor level so it should run at host speed right? -- Adam Vande More
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?AANLkTin6gZiSenKK0MMQm1gfyO_YsV=-b9aDEUR-wKHg>