All of lore.kernel.org
 help / color / mirror / Atom feed
From: xuehai zhang <hai@cs.uchicago.edu>
To: xen-devel@lists.xensource.com
Cc: Tim Freeman <tfreeman@mcs.anl.gov>,
	Kate Keahey <keahey@mcs.anl.gov>,
	xen-usersl@lists.xensource.com
Subject: question about disk performance in domU
Date: Mon, 21 Nov 2005 09:41:08 -0600	[thread overview]
Message-ID: <4381EA94.7060604@cs.uchicago.edu> (raw)

Hi all,

When I ran the experiments to compare an application's execution time in both a domU (named cctest1) 
and a native Linux machine (named ccn10), I noticed the application executes faster in domU. The 
host of the domU (named ccn9) and ccn10 are two nodes of a cluster and they have same hardware 
configurations. domU (cctest1) is created by exporting loopback files from dom0 on ccn9 as VBD 
backends. The application execution logs there might be some disk I/O difference between cctest and 
ccn10, so I did some disk performance profiling with "hdparms" on cctest1 (domU), ccn10 (native 
Linux), ccn9 (dom0), and ccn9 (native Linux). Also, I checked the "DMA" config information from the 
output of dmesg. I tried to run "hdparm -i" and "hdparm -I" but they didn't work. Seems they didn't 
work with SCSI disks. The following are the results.
Thanks in advance for your help.
Best,
Xuehai

1. cctest1 (dumU)
**************************************************************************************************
cctest1$ df -lh
Filesystem            Size  Used Avail Use% Mounted on
/dev/sda1             1.5G  1.1G  306M  78% /
tmpfs                  62M  4.0K   62M   1% /dev/shm
/dev/sda6             4.2G  3.6G  453M  89% /tmp
/dev/sda5             938M  205M  685M  23% /var

cctest1$ dmesg | grep DMA
   DMA zone: 101376 pages, LIFO batch:16

cctest1$ for i in `seq 1 10`; do hdparm -tT /dev/sda1; done

  Timing cached reads:   512 MB in  2.00 seconds = 256.00 MB/sec
  Timing buffered disk reads:   44 MB in  3.00 seconds =  14.67 MB/sec

  Timing cached reads:   528 MB in  2.01 seconds = 262.69 MB/sec
  Timing buffered disk reads:   84 MB in  3.08 seconds =  27.27 MB/sec

  Timing cached reads:   520 MB in  2.00 seconds = 260.00 MB/sec
  Timing buffered disk reads:  120 MB in  3.06 seconds =  39.22 MB/sec

  Timing cached reads:   520 MB in  2.00 seconds = 260.00 MB/sec
  Timing buffered disk reads:  150 MB in  3.06 seconds =  49.02 MB/sec

  Timing cached reads:   536 MB in  2.00 seconds = 268.00 MB/sec
  Timing buffered disk reads:  178 MB in  3.17 seconds =  56.15 MB/sec

  Timing cached reads:   536 MB in  2.00 seconds = 268.00 MB/sec
  Timing buffered disk reads:  204 MB in  3.08 seconds =  66.23 MB/sec

  Timing cached reads:   532 MB in  2.00 seconds = 266.00 MB/sec
  Timing buffered disk reads:  228 MB in  3.13 seconds =  72.84 MB/sec

  Timing cached reads:   540 MB in  2.01 seconds = 268.66 MB/sec
  Timing buffered disk reads:  248 MB in  3.04 seconds =  81.58 MB/sec

  Timing cached reads:   540 MB in  2.00 seconds = 270.00 MB/sec
  Timing buffered disk reads:  266 MB in  3.06 seconds =  86.93 MB/sec

  Timing cached reads:   532 MB in  2.00 seconds = 266.00 MB/sec
  Timing buffered disk reads:  282 MB in  3.06 seconds =  92.16 MB/sec

**************************************************************************************************


2. ccn10 (native Linux)

**************************************************************************************************

ccn10$ df -lh
Filesystem            Size  Used Avail Use% Mounted on
/dev/sda1             1.5G  1.3G  149M  90% /
tmpfs                 252M     0  252M   0% /dev/shm
/dev/sda6             4.2G  3.6G  358M  92% /tmp
/dev/sda5             938M  706M  184M  80% /var

ccn10$ dmesg | grep DMA
   DMA zone: 4096 pages, LIFO batch:1

ccn10$ for i in `seq 1 10`; do hdparm -tT /dev/sda1; done

  Timing cached reads:   516 MB in  2.00 seconds = 257.78 MB/sec
  Timing buffered disk reads:   62 MB in  3.03 seconds =  20.47 MB/sec

  Timing cached reads:   524 MB in  2.01 seconds = 261.00 MB/sec
  Timing buffered disk reads:   62 MB in  3.01 seconds =  20.61 MB/sec

  Timing cached reads:   516 MB in  2.00 seconds = 257.65 MB/sec
  Timing buffered disk reads:   62 MB in  3.01 seconds =  20.61 MB/sec

  Timing cached reads:   524 MB in  2.00 seconds = 262.04 MB/sec
  Timing buffered disk reads:   62 MB in  3.01 seconds =  20.61 MB/sec

  Timing cached reads:   516 MB in  2.00 seconds = 257.78 MB/sec
  Timing buffered disk reads:   62 MB in  3.02 seconds =  20.51 MB/sec

  Timing cached reads:   524 MB in  2.00 seconds = 261.78 MB/sec
  Timing buffered disk reads:   62 MB in  3.02 seconds =  20.52 MB/sec

  Timing cached reads:   516 MB in  2.00 seconds = 257.78 MB/sec
  Timing buffered disk reads:   62 MB in  3.02 seconds =  20.51 MB/sec

  Timing cached reads:   524 MB in  2.00 seconds = 261.78 MB/sec
  Timing buffered disk reads:   62 MB in  3.02 seconds =  20.50 MB/sec

  Timing cached reads:   516 MB in  2.00 seconds = 257.40 MB/sec
  Timing buffered disk reads:   64 MB in  3.09 seconds =  20.73 MB/sec

  Timing cached reads:   524 MB in  2.01 seconds = 260.87 MB/sec
  Timing buffered disk reads:   62 MB in  3.01 seconds =  20.61 MB/sec

**************************************************************************************************


3. ccn9 (dom0)
**************************************************************************************************

ccn9$ df -lh
Filesystem            Size  Used Avail Use% Mounted on
/dev/sda1             1.5G  1.1G  306M  78% /
tmpfs                  62M  4.0K   62M   1% /dev/shm
/dev/sda6             4.2G  3.6G  453M  89% /tmp
/dev/sda5             938M  205M  685M  23% /var

ccn9$ dmesg | grep DMA
   DMA zone: 32768 pages, LIFO batch:8

ccn9$ for i in `seq 1 10`; do hdparm -tT /dev/sda1; done

  Timing cached reads:   504 MB in  2.00 seconds = 252.00 MB/sec
  Timing buffered disk reads:   60 MB in  3.14 seconds =  19.11 MB/sec

  Timing cached reads:   516 MB in  2.00 seconds = 258.00 MB/sec
  Timing buffered disk reads:   62 MB in  3.15 seconds =  19.68 MB/sec

  Timing cached reads:   512 MB in  2.00 seconds = 256.00 MB/sec
  Timing buffered disk reads:   60 MB in  3.08 seconds =  19.48 MB/sec

  Timing cached reads:   516 MB in  2.00 seconds = 258.00 MB/sec
  Timing buffered disk reads:   58 MB in  3.02 seconds =  19.21 MB/sec

  Timing cached reads:   516 MB in  2.01 seconds = 256.72 MB/sec
  Timing buffered disk reads:   60 MB in  3.12 seconds =  19.23 MB/sec

  Timing cached reads:   520 MB in  2.00 seconds = 260.00 MB/sec
  Timing buffered disk reads:   60 MB in  3.13 seconds =  19.17 MB/sec

  Timing cached reads:   520 MB in  2.01 seconds = 258.71 MB/sec
  Timing buffered disk reads:   60 MB in  3.13 seconds =  19.17 MB/sec

  Timing cached reads:   520 MB in  2.01 seconds = 258.71 MB/sec
  Timing buffered disk reads:   60 MB in  3.06 seconds =  19.61 MB/sec

  Timing cached reads:   516 MB in  2.01 seconds = 256.72 MB/sec
  Timing buffered disk reads:   60 MB in  3.14 seconds =  19.11 MB/sec

  Timing cached reads:   516 MB in  2.00 seconds = 258.00 MB/sec
  Timing buffered disk reads:   60 MB in  3.15 seconds =  19.05 MB/sec

**************************************************************************************************

4. ccn9 (native Linux)
**************************************************************************************************
ccn9$ df -lh
Filesystem            Size  Used Avail Use% Mounted on
/dev/sda1             1.5G  1.1G  306M  78% /
tmpfs                  62M  4.0K   62M   1% /dev/shm
/dev/sda6             4.2G  3.6G  453M  89% /tmp
/dev/sda5             938M  205M  685M  23% /var

ccn9 $ dmesg | grep DMA
   DMA zone: 4096 pages, LIFO batch:1

ccn9$ for i in `seq 1 10`; do hdparm -tT /dev/sda1; done
/dev/sda1:
  Timing cached reads:   492 MB in  2.01 seconds = 244.57 MB/sec
  Timing buffered disk reads:   62 MB in  3.10 seconds =  20.01 MB/sec

  Timing cached reads:   484 MB in  2.01 seconds = 241.07 MB/sec
  Timing buffered disk reads:   48 MB in  3.01 seconds =  15.95 MB/sec

  Timing cached reads:   484 MB in  2.00 seconds = 241.67 MB/sec
  Timing buffered disk reads:   62 MB in  3.03 seconds =  20.45 MB/sec

  Timing cached reads:   484 MB in  2.01 seconds = 241.31 MB/sec
  Timing buffered disk reads:   62 MB in  3.01 seconds =  20.57 MB/sec

  Timing cached reads:   480 MB in  2.01 seconds = 239.08 MB/sec
  Timing buffered disk reads:   62 MB in  3.03 seconds =  20.49 MB/sec

  Timing cached reads:   488 MB in  2.01 seconds = 243.31 MB/sec
  Timing buffered disk reads:   62 MB in  3.05 seconds =  20.31 MB/sec

  Timing cached reads:   484 MB in  2.01 seconds = 241.31 MB/sec
  Timing buffered disk reads:   62 MB in  3.01 seconds =  20.61 MB/sec

  Timing cached reads:   484 MB in  2.00 seconds = 241.67 MB/sec
  Timing buffered disk reads:   62 MB in  3.01 seconds =  20.59 MB/sec

  Timing cached reads:   488 MB in  2.01 seconds = 242.34 MB/sec
  Timing buffered disk reads:   62 MB in  3.01 seconds =  20.59 MB/sec

  Timing cached reads:   484 MB in  2.01 seconds = 240.35 MB/sec
  Timing buffered disk reads:   62 MB in  3.09 seconds =  20.09 MB/sec
**************************************************************************************************

             reply	other threads:[~2005-11-21 15:41 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-11-21 15:41 xuehai zhang [this message]
2005-11-21 20:45 ` question about disk performance in domU Tim Freeman
  -- strict thread matches above, loose matches on Subject: below --
2005-11-21 22:26 Santos, Jose Renato G
2005-11-21 22:41 ` xuehai zhang
2005-11-21 22:51   ` Adam Heath
2005-11-21 23:58     ` xuehai zhang

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=4381EA94.7060604@cs.uchicago.edu \
    --to=hai@cs.uchicago.edu \
    --cc=keahey@mcs.anl.gov \
    --cc=tfreeman@mcs.anl.gov \
    --cc=xen-devel@lists.xensource.com \
    --cc=xen-usersl@lists.xensource.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.