<table cellspacing="0" cellpadding="0" border="0" ><tr><td valign="top" style="font: inherit;"><DIV id=yiv240799014>
<TABLE class=yiv240799014 id=yiv240799014bodyDrftID cellSpacing=0 cellPadding=0 border=0>
<TBODY>
<TR>
<TD id=yiv240799014drftMsgContent style="FONT-SIZE: 10pt; FONT-FAMILY: arial">
<DIV>That's so weird, we never got that performance. I just tried your command, it's very slow. The dynamic format is super fast.</DIV>
<DIV> </DIV>
<DIV>VBoxManage.exe createhd --filename c:\test.vhd --size 10240 --format vhd --variant fixed</DIV>
<DIV> </DIV>
<DIV>If I use the following (which creates a file with 10G), then it's very fast.<BR></DIV>
<DIV> </DIV>
<DIV>fsutil file createnew c:\test.raw 10737418240</DIV>
<DIV><BR>--- On <B>Tue, 10/19/10, Klaus Espenlaub <I><klaus.espenlaub@oracle.com></I></B> wrote:<BR></DIV>
<BLOCKQUOTE style="PADDING-LEFT: 5px; MARGIN-LEFT: 5px; BORDER-LEFT: rgb(16,16,255) 2px solid"><BR>From: Klaus Espenlaub <klaus.espenlaub@oracle.com><BR>Subject: Re: [vbox-dev] is virtstor being implemented?<BR>To: vbox-dev@virtualbox.org<BR>Date: Tuesday, October 19, 2010, 3:00 AM<BR><BR>
<DIV class=yiv240799014plainMail>On 19.10.2010 09:15, Huihong Luo wrote:<BR>> Achim,<BR>> Thanks for the clarification.<BR>> I was playing Microsoft Virtual Server 2005 SP1 this morning, and they<BR>> have SCSI Shunt Driver, a kind of virtstor, and it causes quite a<BR>> difference when using it, so I was wondering there might be something.<BR>> Now I know it must be due to their poor implementation on regular disk<BR>> emulation.<BR>> VBox team is superb ! The disk performance is impressive !<BR>> There is one minor thing about fixed size disk creation, it seems to be<BR>> quite slow. 10G VHD fixed size takes about 30 mins. Is this normal?<BR>> Microsoft's VHD driver on windows 7 seems to be much faster. I know this<BR>> isn't very important, as it's not used at runtime.<BR><BR>What? "VHD fixed" creation in VirtualBox is super fast, for 10G it takes <BR>under a second here (VBoxManage createhd /path/to/test.vhd --size
10240 <BR>--format vhd --variant fixed). In fact we have a TODO in the code to <BR>make it slower. At the moment the code just sets the file size to the <BR>required capacity, making it a sparse file. We made VERY bad experience <BR>with sparse file performance on Windows, writing a single 512 byte block <BR>to such a file on NTFS can take several minutes, which might be your <BR>real problem...<BR><BR>We don't really consider VHD a very important image format. The code is <BR>well tested and has generally good performance, however it hasn't seen <BR>much optimization. I personally wouldn't use it unless I absolutely have <BR>to, and the only reason would be existing images.<BR><BR>Klaus<BR><BR>><BR>><BR>> --- On *Mon, 10/18/10, Achim Hasenmüller<BR>> /<<A href="http://us.mc343.mail.yahoo.com/mc/compose?to=achim.hasenmueller@oracle.com" target=_blank rel=nofollow>achim.hasenmueller@oracle.com</A>>/*
wrote:<BR>><BR>><BR>> From: Achim Hasenmüller <<A href="http://us.mc343.mail.yahoo.com/mc/compose?to=achim.hasenmueller@oracle.com" target=_blank rel=nofollow>achim.hasenmueller@oracle.com</A>><BR>> Subject: Re: [vbox-dev] is virtstor being implemented?<BR>> To: "Huihong Luo" <<A href="http://us.mc343.mail.yahoo.com/mc/compose?to=huisinro@yahoo.com" target=_blank rel=nofollow>huisinro@yahoo.com</A>><BR>> Cc: "Alexey Eromenko" <<A href="http://us.mc343.mail.yahoo.com/mc/compose?to=al4321@gmail.com" target=_blank rel=nofollow>al4321@gmail.com</A>>, "Liang Suilong"<BR>> <<A href="http://us.mc343.mail.yahoo.com/mc/compose?to=liangsuilong@gmail.com" target=_blank rel=nofollow>liangsuilong@gmail.com</A>>, <A href="http://us.mc343.mail.yahoo.com/mc/compose?to=vbox-dev@virtualbox.org" target=_blank
rel=nofollow>vbox-dev@virtualbox.org</A><BR>> Date: Monday, October 18, 2010, 11:57 PM<BR>><BR>> What you describe is basically how SATA and SCSI/SAS work :-)<BR>><BR>> Honestly, we don't see any reason to implement virtstor. We do not<BR>> expect any performance advantages and whereas a network driver is<BR>> easy to add/change after install, a disk controller is more<BR>> complicated and less user friendly.<BR>><BR>> I would say that there isn't much out there that can compete with<BR>> VBox when it comes to disk performance. Our performance work is now<BR>> mostly focused around network as this is clearly where we have room<BR>> for improvement. We want to continue to improve
the E1000 device but<BR>> for ultimate performance, our efforts are currently focused on<BR>> virtio. Once we can make full use of receive and send offloading, we<BR>> should be good.<BR>><BR>> Achim<BR>><BR>> On Oct 19, 2010, at 8:14 , Huihong Luo wrote:<BR>><BR>>> yes, it's the speed reason. In theory, virtstor is supposed to be<BR>>> fast. However, since vbox is already doing so well on disk, I am<BR>>> not sure how much further can be improved.<BR>>> In theory, virtstor bypasses many calls, and goes directly to the<BR>>> cheese, i.e., block i/o data transfer. The performance depends how<BR>>> well the mechanism for the transfer
between vm and host, which is<BR>>> why I mentioned share folders. (btw, I know the difference between<BR>>> file sysem and block driver, I do that for a living). What I mean<BR>>> is that we can use the same large data transfer mechanism if<BR>>> shared folder's performance is very good.<BR>>> for example, a user app wants to write 10k data to the disk, it<BR>>> goes to the virtio driver after a few calls in windows, if the<BR>>> virtstor driver can efficiently (directly) transfer the buffer to<BR>>> the host, perhaps through memory sharing, or pinned physical<BR>>> pages, this should improve the performance.<BR>>> About the guest drivers, windows 7 and
vista sp1 introduced some<BR>>> new models that should be even faster than kvm's implementation.<BR>>><BR>>><BR>><BR>><BR>><BR>> _______________________________________________<BR>> vbox-dev mailing list<BR>> <A href="http://us.mc343.mail.yahoo.com/mc/compose?to=vbox-dev@virtualbox.org" target=_blank rel=nofollow>vbox-dev@virtualbox.org</A><BR>> <A href="http://vbox.innotek.de/mailman/listinfo/vbox-dev" target=_blank rel=nofollow>http://vbox.innotek.de/mailman/listinfo/vbox-dev</A><BR><BR><BR>_______________________________________________<BR>vbox-dev mailing list<BR><A href="http://us.mc343.mail.yahoo.com/mc/compose?to=vbox-dev@virtualbox.org" target=_blank rel=nofollow>vbox-dev@virtualbox.org</A><BR><A href="http://vbox.innotek.de/mailman/listinfo/vbox-dev" target=_blank
rel=nofollow>http://vbox.innotek.de/mailman/listinfo/vbox-dev</A><BR></DIV></BLOCKQUOTE></TD></TR></TBODY></TABLE></DIV></td></tr></table>