Results 1 to 20 of 20
  1. #1
    Join Date
    Sep 2006
    Location
    Toronto
    Posts
    163

    * How to test I/O performance without using the famous dd command

    few days ago I have signed up a managed vps. After optimizing and initial setup, I ran the dd command to test the I/O performance:

    dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync

    and it wasn't impressed at all, I never get anything higher than 3xmb/s, most of the time I got 2xmb/s, and sometimes 1xmb/s

    I have contacted my vps provider and told them the situation, either they tried to convince me by running:

    dd if=/dev/zero of=test bs=1M count=1024
    (without conv=fdatasync)
    (for those who want to know the difference with or without conv=fdatasync please see this article http://www.lowendbox.com/blog/hostra...#comment-14744)

    or telling me there is no problem, server is not overloaded etc and asked me if my vps is sluggish or not. There's not much help from them

    At this moment I still haven't migrated any of my website to this new vps, as I want to make sure everything is fine to do so. If not, I'm not gonna keep it for another month. So besides using the dd command, is there anything I can run to test the disk performance or check if the node is overloaded. I'm running openvz w/centos 6.

    Thanks in advance

  2. #2
    Join Date
    Oct 2006
    Location
    State College, PA USA
    Posts
    370
    Quote Originally Posted by Red Jersey View Post
    I ran the dd command to test the I/O performance:

    dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync

    and it wasn't impressed at all, I never get anything higher than 3xmb/s, most of the time I got 2xmb/s, and sometimes 1xmb/s
    Please be careful with your units. A bit is not equal to a byte and the units are abbreviated with very specific capitalization so that we can all interpret them properly. 10MB/s is not the same as 10mb/s. Are you saying that you got 3mb/s as a result from dd? As in three mega BITS per second? Please copy/past the output directly from dd without modifying it and we'll be able to interpret it more accurately.

    Quote Originally Posted by Red Jersey View Post
    I have contacted my vps provider and told them the situation, either they tried to convince me by running:

    dd if=/dev/zero of=test bs=1M count=1024
    What are the results using this dd syntax? Even a regular old SATA hard drive should be able to easily achieve 10MB/s. It does sound like there's something not quite right with this situation.
    Cheap VPS Hosting with minimum 99.9% SLA.
    State College, PA VPS Hosting

  3. #3
    Join Date
    Sep 2006
    Location
    Toronto
    Posts
    163
    sorry, when i said 2xmb/s or 1xmb/s I meant the numbers betwen 20-29 or 10-19

    here is what i got:
    dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync
    16384+0 records in
    16384+0 records out
    1073741824 bytes (1.1 GB) copied, 47.6071 s, 22.6 MB/s

    sorry if it confused you

  4. #4
    Join Date
    Oct 2006
    Location
    State College, PA USA
    Posts
    370
    Quote Originally Posted by Red Jersey View Post
    sorry, when i said 2xmb/s or 1xmb/s I meant the numbers betwen 20-29 or 30-39

    sorry if it confused you
    Ahh now I understand. I guess I should have seen the x in there. Sorry about that!

    Quote Originally Posted by Red Jersey View Post
    here is what i got:
    dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync
    16384+0 records in
    16384+0 records out
    1073741824 bytes (1.1 GB) copied, 47.6071 s, 22.6 MB/s
    That's not GREAT, but it's not terrible. I think for most the low-end VPS providers, that is pretty much the going average. If that's not good enough and you ARE on a low-end provider, it's probably time to move up. If you are on a HIGH-end provider, or even a middle of the road provider, you are very correct in escalating this to their support. I hope you get a resolution soon!

    Best,
    Dave
    Cheap VPS Hosting with minimum 99.9% SLA.
    State College, PA VPS Hosting

  5. #5
    Join Date
    Nov 2010
    Posts
    232
    When you next send in a support ticket/reply about this, give them a link to this thread so they can learn a few things.

  6. #6
    Join Date
    May 2001
    Location
    HK
    Posts
    3,082

    *

    Quote Originally Posted by Red Jersey View Post
    sorry, when i said 2xmb/s or 1xmb/s I meant the numbers betwen 20-29 or 10-19

    here is what i got:
    dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync
    16384+0 records in
    16384+0 records out
    1073741824 bytes (1.1 GB) copied, 47.6071 s, 22.6 MB/s

    sorry if it confused you
    What number impresses you? And how often would you have to do a sequential write to a large file?

    If it does not affect the operation of your server, why bother? You might get better performance when you move on to the next provider, but doesn't mean the performance will last forever. Things will turn slower and slower gradually as soon as more VMs are running.

  7. #7
    root@vps [~]# dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync
    16384+0 records in
    16384+0 records out
    1073741824 bytes (1.1 GB) copied, 7.61077 seconds, 141 MB/s
    VPS from future hosting

  8. #8
    Join Date
    Jul 2010
    Location
    ~/
    Posts
    1,382
    dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync
    16384+0 records in
    16384+0 records out
    1073741824 bytes (1.1 GB) copied, 3.59642 seconds, 299 MB/s
    From our test rig running 5+ year old sata drives in software raid... my point.... the test is completely invalid.

    As already said unless you plan to do allot of sequential writes to a large file the dd test is meaningless however you did ask for a way to test IO performance so before this thread turns in to a "my balls are bigger than yours dd thread"

    You should be doing a number of tests to benchmark your whole VPS not just the disk.

    Unixbench
    Geekbench
    ioping (very important for disk) http://code.google.com/p/ioping/

    Is your VPS on 1000 or 100mbit if 1000 mbit can you download files faster that the dd test can write, if so consider the logic
    -> INCEPTION HOSTING LIMITED Since 2010!
    -> I am most active on the lowendspirit hosting forum Come join us!
    -> PHOENIX USA & THE NETHERLANDS & UK EU

  9. #9
    Join Date
    Jun 2010
    Location
    Northern Ireland
    Posts
    45
    I'm getting 87.6MB/s on Linode and 77.1MB/s on Burstnet (low end) server. Just throwing that in for comparison.

  10. #10
    Join Date
    Jun 2006
    Posts
    1,767
    This is not an accurate test of IO speed.

    Please try hdparm -t /dev/sda.

  11. #11
    Join Date
    Oct 2006
    Location
    State College, PA USA
    Posts
    370
    Quote Originally Posted by backtogeek View Post
    before this thread turns in to a "my balls are bigger than yours dd thread"
    Haha. Couldn't have said it better myself!
    Cheap VPS Hosting with minimum 99.9% SLA.
    State College, PA VPS Hosting

  12. #12
    Join Date
    Dec 2007
    Posts
    471
    Also try looking at ioping to test disk latency.

  13. #13
    Quote Originally Posted by backtogeek View Post
    ioping (very important for disk)
    Could you tell me how to use ioping properly?

  14. #14
    Join Date
    Jun 2011
    Location
    California
    Posts
    412
    There's a good amount of information here: http://code.google.com/p/ioping/

    If you're using BSD or another UNIX operating system, change the devices accordingly.

    Code:
    Usage: ioping [-LCDRq] [-c count] [-w deadline] [-p period] [-i interval]
                   [-s size] [-S wsize] [-o offset] device|file|directory
            ioping -h | -v
    
          -c <count>      stop after <count> requests
          -w <deadline>   stop after <deadline>
          -p <period>     print raw statistics for every <period> requests
          -i <interval>   interval between requests (1s)
          -s <size>       request size (4k)
          -S <wsize>      working set size (1m)
          -o <offset>     in file offset
          -L              use sequential operations (includes -s 256k)
          -C              use cached I/O
          -D              use direct I/O
          -R              seek rate test (same as -q -i 0 -w 3 -S 64m)
          -q              suppress human-readable output
          -h              display this message and exit
          -v              display version and exit
    Show disk I/O latency using the default values and the current directory, until interrupted

    Code:
    $ ioping .
    4096 bytes from . (ext4 /dev/sda3): request=1 time=0.2 ms
    4096 bytes from . (ext4 /dev/sda3): request=2 time=0.2 ms
    4096 bytes from . (ext4 /dev/sda3): request=3 time=0.3 ms
    4096 bytes from . (ext4 /dev/sda3): request=4 time=12.7 ms
    4096 bytes from . (ext4 /dev/sda3): request=5 time=0.3 ms
    ^C
    --- . (ext4 /dev/sda3) ioping statistics ---
    5 requests completed in 4794.0 ms, 364 iops, 1.4 mb/s
    min/avg/max/mdev = 0.2/2.8/12.7/5.0 ms
    Measure disk seek rate (iops, avg)
    $ ioping -R /dev/sda

    Code:
    --- /dev/sda (device 465.8 Gb) ioping statistics ---
    186 requests completed in 3004.6 ms, 62 iops, 0.2 mb/s
    min/avg/max/mdev = 6.4/16.0/26.8/4.7 ms
    Measure disk sequential speed (mb/s)
    $ ioping -RL /dev/sda

    Code:
    --- /dev/sda (device 465.8 Gb) ioping statistics ---
    837 requests completed in 3004.1 ms, 292 iops, 72.9 mb/s
    min/avg/max/mdev = 2.0/3.4/28.9/2.0 ms
    |███| SSD Nodes
    |███| Build more for far less with the best value in cloud hosting.
    |███| RAID 10 SSDs | 10Gbps NICs | Tier-1 backbone | Scale without downtime | 9 years in business
    |███| https://www.ssdnodes.com

  15. #15
    Join Date
    Jul 2010
    Location
    ~/
    Posts
    1,382
    ^^ What he said ^^

    Quote Originally Posted by tdc-adm View Post
    Could you tell me how to use ioping properly?
    -> INCEPTION HOSTING LIMITED Since 2010!
    -> I am most active on the lowendspirit hosting forum Come join us!
    -> PHOENIX USA & THE NETHERLANDS & UK EU

  16. #16
    Join Date
    Sep 2006
    Location
    Toronto
    Posts
    163
    so here is what i got:

    ioping -c 10 /
    4096 bytes from / (simfs /dev/simfs): request=1 time=0.1 ms
    4096 bytes from / (simfs /dev/simfs): request=2 time=0.1 ms
    4096 bytes from / (simfs /dev/simfs): request=3 time=0.1 ms
    4096 bytes from / (simfs /dev/simfs): request=4 time=0.1 ms
    4096 bytes from / (simfs /dev/simfs): request=5 time=3708.7 ms
    4096 bytes from / (simfs /dev/simfs): request=6 time=0.1 ms
    4096 bytes from / (simfs /dev/simfs): request=7 time=1033.5 ms
    4096 bytes from / (simfs /dev/simfs): request=8 time=447.4 ms
    4096 bytes from / (simfs /dev/simfs): request=9 time=110.7 ms
    4096 bytes from / (simfs /dev/simfs): request=10 time=11.1 ms

  17. #17
    Join Date
    Jul 2010
    Location
    ~/
    Posts
    1,382
    0.1 = fine
    0.1 = fine
    0.1 = fine
    0.1 = fine
    3708.7 = Jesus H!!
    0.1 = fine
    etc etc.

    The odd spike is not unusual, someone else may have been giving it dd's at the time, for a real test you should be looking at 1000 minimum

    If you cba to wait for it just echo the results to a file with '> ioping.txt &' at the end of the command and check the results when its finished.

    Again the odd spike is normal but if you start getting, I dunno 10 - 20 nasty results in a row just sent the output to your host and ask what's going on, I suspect they wont have an answer though based on the previous info you gave.

    Would be interesting to see your unix or geek bench scores too just as general reference.
    -> INCEPTION HOSTING LIMITED Since 2010!
    -> I am most active on the lowendspirit hosting forum Come join us!
    -> PHOENIX USA & THE NETHERLANDS & UK EU

  18. #18
    Join Date
    Sep 2006
    Location
    Toronto
    Posts
    163
    my ioping result:

    PHP Code:
    tail -f ioping.result
    4096 bytes from 
    / (simfs /dev/simfs): request=995 time=0.1 ms
    4096 bytes from 
    / (simfs /dev/simfs): request=996 time=0.2 ms
    4096 bytes from 
    / (simfs /dev/simfs): request=997 time=1293.9 ms
    4096 bytes from 
    / (simfs /dev/simfs): request=998 time=31.0 ms
    4096 bytes from 
    / (simfs /dev/simfs): request=999 time=0.2 ms
    4096 bytes from 
    / (simfs /dev/simfs): request=1000 time=0.2 ms

    --- / (simfs /dev/simfsioping statistics ---
    1000 requests completed in 1073183.8 ms14 iops0.1 mb/s
    min
    /avg/max/mdev 0.0/73.1/7576.7/491.7 ms 
    geekbench:
    http://browse.geekbench.ca/geekbench2/view?id=469968

    I paid $25+ for this, not sure if I should keep it or not, please advise

  19. #19
    Join Date
    Jul 2010
    Location
    ~/
    Posts
    1,382
    The GB score is actually pretty decent, and if you mean $25 p/year then its very good (compare this score with some of the ones on http://96mb.com for reference)

    However.... your abv result in the ioping result is not good, if you can work with your provider to work this out then its a keeper I would say.

    I suppose the bottom line though.... apart from the perceived low performance from dd (which is not all that relevant) is it actually affecting you and is anything suffering, if not then its not the end of the world.
    -> INCEPTION HOSTING LIMITED Since 2010!
    -> I am most active on the lowendspirit hosting forum Come join us!
    -> PHOENIX USA & THE NETHERLANDS & UK EU

  20. #20
    Join Date
    Sep 2006
    Location
    Toronto
    Posts
    163
    Quote Originally Posted by backtogeek View Post
    The GB score is actually pretty decent, and if you mean $25 p/year then its very good (compare this score with some of the ones on http://96mb.com for reference)

    However.... your abv result in the ioping result is not good, if you can work with your provider to work this out then its a keeper I would say.

    I suppose the bottom line though.... apart from the perceived low performance from dd (which is not all that relevant) is it actually affecting you and is anything suffering, if not then its not the end of the world.
    of course not, i paid $25+/mo

    I have already contacted them regarding the disk, I even sent them the dd result that I got (around 20ish mb/s) it just they keep saying there is no problem...

    Anyways will try to send the latest ioping result, if they are still in denial then I'll ask for a refund and post my honest review on wht.

    Anyways thanks backtogeek

Similar Threads

  1. How to test performance from China?
    By shunter1 in forum Hosting Security and Technology
    Replies: 5
    Last Post: 10-12-2010, 11:53 PM
  2. any command to test HDD health on server ?
    By ttgt in forum Hosting Security and Technology
    Replies: 6
    Last Post: 11-24-2009, 11:17 AM
  3. help to test performance
    By gate2vn in forum Other Reviews
    Replies: 18
    Last Post: 01-25-2008, 05:25 PM
  4. What command detects apache performance?
    By Emin in forum Hosting Security and Technology
    Replies: 7
    Last Post: 05-24-2005, 01:35 AM
  5. Any performance test?
    By phptalk in forum Dedicated Server
    Replies: 3
    Last Post: 04-10-2002, 08:37 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •