From: Mathias Klette <mklette@gmail.com>
To: Stefan Hajnoczi <stefanha@gmail.com>
Cc: kvm@vger.kernel.org
Subject: Re: Degraded performance with Windows 2008 R2 with applications
Date: Mon, 07 Mar 2011 19:10:41 +0100 [thread overview]
Message-ID: <4D751FA1.6020205@gmail.com> (raw)
In-Reply-To: <AANLkTi=FAePaXFtVotnfArNPCBpV0Y4zBOv93xdSERZC@mail.gmail.com>
On 07/03/11 10:41, Stefan Hajnoczi wrote:
> On Sun, Mar 6, 2011 at 10:25 PM, Mathias Klette<mklette@gmail.com> wrote:
>> I've tested with iozone to compare IO with a linux guest and also to
>> verify changes made to improve situation - but nothing really helped.
>>
>> TESTS with "iozone -s 4G -r 256k -c -e":
>
> Please use the -I option to bypass the page cache, otherwise buffered
> I/O will be used and requests may be satisfied from memory rather than
> actually accessing the disk.
>
thanks, wasn't really aware of that switch.
here now tests, with a raw image on ext4 LV mounted with option
relatime. vm has 4cores, 8G memory.
random
random bkwd record stride
KB reclen write rewrite read reread read
write read rewrite read fwrite frewrite fread freread
4194304 256 116922 43874 53701 34459 62807
40152 152029 3519190 126013 38504 74360 24814 25112
4194304 256 102275 48545 24342 33265 67412
40945 59097 3500722 114174 73353 44401 46383 31860
4194304 256 49306 45876 31610 31285 187091
37275 59135 3472783 54016 65711 70700 24674 49231
4194304 256 85590 90679 31319 47120 269864
38727 137024 3462054 48674 73425 44879 46736 31684
4194304 256 40474 27725 46137 46084 180579
38938 57869 3473130 52202 36281 90943 31189 32703
unfortunately kvm_stat wasn't possible as debugfs is not compiled in.
vmstat 15 on host for the time of one iozone run:
procs -----------memory---------- ---swap-- -----io---- -system--
----cpu----
r b swpd free buff cache si so bi bo in cs us sy
id wa
0 0 0 53722472 809620 160285216 0 0 2 17 1 0
0 1 98 0
7 0 0 53726724 809648 160285280 0 0 0 24761 12731
16822 1 1 98 0
7 0 0 53722752 809672 160285296 0 0 0 120753 23707
15666 1 2 97 0
6 0 0 53724780 809688 160285344 0 0 0 86676 13572
17537 1 1 98 0
2 0 0 53725188 809716 160285392 0 0 1 72324 23238
16397 0 2 98 0
1 0 0 53728600 809740 160285424 0 0 0 24 10909
13529 0 1 99 0
0 0 0 53725612 809760 160285488 0 0 1 44 10469
12467 0 1 99 0
0 0 0 53726592 809784 160285680 0 0 1 299 11871
15643 0 1 99 0
1 0 0 53727512 809800 160285696 0 0 0 29 10635
13284 0 1 99 0
2 0 0 53727952 809832 160285776 0 0 1 53 10483
12935 0 1 99 0
1 0 0 53726348 809856 160285840 0 0 1 54 10384
12600 0 1 99 0
1 0 0 53727096 809880 160285856 0 0 0 108 10902
13659 0 1 99 0
1 0 0 53727328 809904 160285920 0 0 0 38 10348
12692 0 1 99 0
0 0 0 53728328 809928 160285952 0 0 0 33 10137
12195 0 1 99 0
1 0 0 53727680 809952 160286000 0 0 0 35 10653
13327 0 1 99 0
1 0 0 53727616 809968 160286112 0 0 0 31 10465
12932 0 0 99 0
0 0 0 53728148 809992 160286128 0 0 0 17 10037
11946 0 0 99 0
0 0 0 53727756 810016 160286192 0 0 1 29 13424
18647 1 1 98 0
5 0 0 53711500 810040 160286224 0 0 0 41 16580
24443 1 2 97 0
3 0 0 53724776 810076 160286288 0 0 1 61 13441
18676 1 1 98 0
0 0 0 53721816 810108 160286320 0 0 0 201432 14520
14391 0 2 97 1
6 0 0 53724704 810128 160286368 0 0 0 36 11088
14476 0 1 99 0
0 0 0 53725068 810164 160286432 0 0 0 78247 11450
12167 1 2 97 1
7 0 0 53626104 810188 160382512 0 0 0 35 11459
14319 1 1 98 0
3 0 0 53605924 810208 160402400 0 0 0 18 12609
15837 2 1 97 0
2 0 0 53589232 810356 160416864 0 0 0 8471 12203
14956 2 1 97 0
3 0 0 53570564 810372 160434128 0 0 0 26 11359
13451 2 1 97 0
0 0 0 53569516 810428 160438400 0 0 0 1735 10793
12836 2 1 98 0
1 0 0 53570652 810452 160438432 0 0 0 24 10816
13483 0 1 99 0
1 0 0 53570472 810476 160438496 0 0 0 61 11613
14829 0 1 99 0
5 0 0 53567268 810500 160438512 0 0 0 22 12978
17810 1 1 98 0
8 0 0 53564892 810528 160438816 0 0 0 150075 25535
23496 0 2 98 0
6 0 0 53565608 810552 160438848 0 0 0 27318 13518
16144 0 1 99 0
4 0 0 53566828 810576 160438912 0 0 0 102318 13783
15730 1 1 98 0
1 0 0 53564064 810600 160438944 0 0 0 84140 18573
20191 1 2 98 0
3 0 0 53564344 810616 160438992 0 0 0 131089 21789
21013 0 2 98 0
6 0 0 53565948 810640 160439008 0 0 0 17110 12010
11942 0 1 99 0
1 0 0 53567380 810656 160439056 0 0 0 47398 18252
13303 0 1 99 0
1 0 0 53550740 810684 160439104 0 0 0 25 10505
12964 0 1 99 0
2 0 0 53564948 810716 160439216 0 0 1 60 10113
12037 0 1 99 0
2 0 0 53566988 810740 160439264 0 0 0 39 10720
13451 0 1 99 0
1 0 0 53566420 810772 160439296 0 0 0 30 10526
13053 0 1 99 0
regards top, i'm afraid to attach a log for the whole time iozone is
being run. only remarkable: once iozone starts writing the kvm process
requests more cpu time and ksoftirqd/18 and flush-253:1 show up for
short times in the top 5 of the list.
> What is the qemu-kvm command-line (ps aux | grep kvm)?
106 14527 28.5 1.5 4413272 4223720 ? Sl 15:04 58:50
/usr/bin/kvm -S -M pc-0.13 -enable-kvm -m 4096 -smp
4,sockets=4,cores=1,threads=1 -name Profitbricks-Linuxtest -uuid
e4594cb9-6958-d564-8ed0-74ba828b9326 -nodefconfig -nodefaults -chardev
socket,id=monitor,path=/var/lib/libvirt/qemu/VM-34.monitor,server,nowait
-mon chardev=monitor,mode=readline -rtc base=localtime -boot dc -drive
file=/images/VM-34.raw,if=none,id=drive-virtio-disk0,boot=on,format=raw,cache=writeback
-device
virtio-blk-pci,bus=pci.0,addr=0x5,drive=drive-virtio-disk0,id=virtio-disk0
-drive
file=/images/virtio-win-1.1.16.iso,if=none,media=cdrom,id=drive-ide0-1-0,readonly=on,format=raw
-device ide-drive,bus=ide.1,unit=0,drive=drive-ide0-1-0,id=ide0-1-0
-netdev tap,fd=90,id=hostnet0 -device
e1000,netdev=hostnet0,id=net0,mac=52:54:00:4e:3c:6d,bus=pci.0,addr=0x3
-netdev tap,fd=95,id=hostnet1 -device
virtio-net-pci,netdev=hostnet1,id=net1,mac=52:54:00:37:4f:5f,bus=pci.0,addr=0x6
-chardev pty,id=serial0 -device isa-serial,chardev=serial0 -usb -device
usb-tablet,id=input0 -vnc 0.0.0.0:10 -k de -vga std -device
virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x4
>
> Are you using virtio-blk and the Windows guest drivers from here:
>
> http://www.linux-kvm.org/page/WindowsGuestDrivers/Download_Drivers
yes, versions 1.1.16. as far as I've seen, there were no newer ones
available?!
>
> Stefan
I checked with linux-2.6.38-rc6 with AMD IOMMU support today and got far
better results and also it felt like a usable system:
random
random bkwd record stride
KB reclen write rewrite read reread read
write read rewrite read fwrite frewrite fread freread
4194304 256 64937 101079 245510 229593 265388
79120 321497 3608593 273455 48558 99760 224877 224175
4194304 256 49674 95625 66813 213311 70686
79345 322612 3198246 272728 96646 99040 212715 210044
4194304 256 97412 49211 204895 654454 246541
75667 74149 3250535 260699 48806 99973 225283 225453
next prev parent reply other threads:[~2011-03-07 18:10 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-03-06 22:25 Degraded performance with Windows 2008 R2 with applications Mathias Klette
2011-03-07 9:41 ` Stefan Hajnoczi
2011-03-07 18:10 ` Mathias Klette [this message]
2011-03-07 9:57 ` Avi Kivity
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4D751FA1.6020205@gmail.com \
--to=mklette@gmail.com \
--cc=kvm@vger.kernel.org \
--cc=stefanha@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox