Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


DD Test Results
New on LowEndTalk? Please Register and read our Community Rules.

All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.

DD Test Results

zonzon Member
edited November 2012 in General

I created a script for ddtest.. and here is the results.. hope this will help somebody...

inceptionhosting_nl
1073741824 bytes (1.1 GB) copied, 19.4186 s, 55.3 MB/s
quickweb_germany
1073741824 bytes (1.1 GB) copied, 11.0016 s, 97.6 MB/s
vpsnine_singapore
1073741824 bytes (1.1 GB) copied, 19.8181 s, 54.2 MB/s
chicagovps
1073741824 bytes (1.1 GB) copied, 8.10727 s, 132 MB/s
buyvmkvm
1073741824 bytes (1.1 GB) copied, 7.69543 s, 140 MB/s
buyvm256
1073741824 bytes (1.1 GB) copied, 12.8746 s, 83.4 MB/s
buyvm128
1073741824 bytes (1.1 GB) copied, 4.41738 s, 243 MB/s
hostigation256
1073741824 bytes (1.1 GB) copied, 7.62545 s, 141 MB/s
hostigationkvm128
1073741824 bytes (1.1 GB) copied, 20.4417 s, 52.5 MB/s
securedragon
1073741824 bytes (1.1 GB) copied, 8.40035 s, 128 MB/s
minivps
1073741824 bytes (1.1 GB) copied, 9.02412 s, 119 MB/s
prometeus
1073741824 bytes (1.1 GB) copied, 3.1842 s, 337 MB/s
budgetvm
1073741824 bytes (1.1 GB) copied, 4.44092 s, 242 MB/s
ssdvm
1073741824 bytes (1.1 GB) copied, 5.5201 s, 195 MB/s
gationftp
1073741824 bytes (1.1 GB) copied, 5.07256 s, 212 MB/s
alienlayer
1073741824 bytes (1.1 GB) copied, 21.7584 s, 49.3 MB/s
ipxcore256
1073741824 bytes (1.1 GB) copied, 14.9491 s, 71.8 MB/s
ipxcore96
1073741824 bytes (1.1 GB) copied, 7.10254 s, 151 MB/s

«134

Comments

  • Nick_ANick_A Member, Top Host, Host Rep

    1 - Need to see the actual commands to know anything about these tests.
    2 - SSDVM is closing.
    3 - Server.Bear.

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

  • thanks OP I came buckets

  • MikeInMikeIn Member
    edited November 2012

    `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, 4.87684 s, 220 MB/s

    `
    Best result in my 192 MB OVZ out of 3 tries.

    Finally the winner is @prometeus

  • MikeInMikeIn Member
    edited November 2012

    Avg:- 205.3333333333333 MB/s @prometeus
    Total tries: 3

  • prometeusprometeus Member, Host Rep

    @MikeIn said: Finally the winner is @prometeus

    it seems my fake dd binary is doing a great job! :P

  • This is the results in VZSSD7 Prometeus :

    # dd if=/dev/zero of=iotest bs=64k count=16k conv=fdatasync && rm -rf iotest
    16384+0 records in
    16384+0 records out
    1073741824 bytes (1.1 GB) copied, 1.51978 s, 707 MB/s

    In VZ3, it's about 200MB/s

  • @ErawanArifNugroho
    Got 707 MB/s in VZSSD7.

    @prometeus said: it seems my fake dd binary is doing a great job! :P

    O_o

  • Yes, the VZSSD in the previous offer, before black friday :D

  • OneAsiaHost KVMSSD-512 plan:

    dd if=/dev/zero of=iotest bs=64k count=16k conv=fdatasync && rm -rf iotest
    16384+0 records in
    16384+0 records out
    1073741824 bytes (1.1 GB) copied, 0.807644 s, 1.3 GB/s
    
  • 
    root@gtfo:~# dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync; unlink test
    16384+0 records in
    16384+0 records out
    1073741824 bytes (1.1 GB) copied, 13.983 s, 76.8 MB/s
    
    

    Where did my 300MB/s go ;_;

    Y U DO DIS TO ME @Francisco

  • FranciscoFrancisco Top Host, Host Rep, Veteran

    @Wintereise said: Where did my 300MB/s go ;_;

    Y U DO DIS TO ME @Francisco

    Ticket and we'll see?

    We've got some upgrades coming for December in SJ so that should clean up that all :)

    Francisco

  • WintereiseWintereise Member
    edited November 2012

    Nah, I'm fine with it. For all it runs, it'd probably be fine with 10M/s, even (Hurrdurr IRC Bouncer)

    Just complaining for no reason other than anything :P

    ( @Francisco )

  • AnthonySmithAnthonySmith Member, Patron Provider

    I am now more convinced that ever that people just buy VPSes to run dd tests.

    Inception:

    node10 ~]# 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, 3.57729 s, 300 MB/s
    

    :)

    It is a point in time test on a shared resource environment using sequential writes only.

    Will the madness ever end?

  • Cloud Shards:

    dd if=/dev/zero of=testfilex bs=64k count=16k conv=fdatasync
    16384+0 records in
    16384+0 records out
    1073741824 bytes (1.1 GB) copied, 1.99437 s, 538 MB/s`

    http://www.lowendtalk.com/discussion/5815/cloudshards-1gb-ssd-cached-45gb-600gb-openvz-benchmark

    Not by me :)

  • AnthonySmithAnthonySmith Member, Patron Provider
    node10 ~]# hdparm -tT /dev/sda
    
    /dev/sda:
     Timing cached reads:   12356 MB in  1.99 seconds = 6220.24 MB/sec
     Timing buffered disk reads:  1168 MB in  3.00 seconds = 388.87 MB/sec
    

    Next big thing? :p

  • @AnthonySmith said: It is a point in time test on a shared resource environment using sequential writes only.

    Will the madness ever end?

    Yup, not sure why peeps don't get this. Its an indicator but a single (or at best typically 3 indicators) rather than performance over time. That's why I like what @serverbear is trying to do with his trended testing.

  • Nick_ANick_A Member, Top Host, Host Rep

    @AnthonySmith said: I am now more convinced that ever that people just buy VPSes to run dd tests.

    Yep.

    This thread is really silly, and that's coming from someone who puts a lot of money into having good DD results.

  • MaouniqueMaounique Host Rep, Veteran

    @craigb said: Yup, not sure why peeps don't get this.

    Probably they wish to bypass the cache since most providers only activate it for reads.
    I never understood this thing, for me 30 mb is the limit where i start to see delays in stuff and is the lowest acceptable limit for a payed VPS, but that varies. Some providers (of free VPSes) are doing fine with only 10 mb, i guess it depends more on the read speed and this is normal since 1/5 of all IOPS are usually writes, the others are reads. ioping gives a much better idea of how oversold the IOPS are.

  • AnthonySmithAnthonySmith Member, Patron Provider
    edited November 2012

    I don't object I suppose as everyone has to learn at some point, but the picture in my head goes along these lines:

    1) Buy VPS
    2) Woohooo 400 MB/s sequential writes on dd test
    3) Install highly optimised LNMP or LLMP stack using someone else's script
    4) From that point on everything I do is based on having good read speeds which I never tested and I will never ever run anything ever that requires fast sequential write speeds from step 1

    herp.

    :)

  • jarjar Patron Provider, Top Host, Veteran
    edited November 2012

    Is this advertising time?

    Dallas: 1073741824 bytes (1.1 GB) copied, 4.1525 s, 259 MB/s
    Lenoir: 1073741824 bytes (1.1 GB) copied, 4.95731 s, 217 MB/s
    Denver: 1073741824 bytes (1.1 GB) copied, 4.97165 s, 216 MB/s

    ;)

    People want good dd test results. They may not need them, but the people have spoken, so it's obviously a strong marketing point.

  • MaouniqueMaounique Host Rep, Veteran
    edited November 2012

    While Prometeus seems to be ahead in this thing, it does not mean we shouldnt explain things.
    OK, the customer has spoken, no problem for this, if they want it, they will get it, but they must know that asking things they dont need will only drive prices up for a gain that is not noticeable in daily usage.
    For the storage plans I am putting up, the dd with 192 MB ram is over 100, for a storage plan the dd test is actually more relevant, however, the network will never saturate that write speed, even if the port would be 1 gbps. So, unless you are copying stuff from one folder to another, that is a resource that will never be used.

  • Hmm... probably someone should start testing memory speed by writing into /dev/shm as well.

  • AnthonySmithAnthonySmith Member, Patron Provider

    @jarland if that is your opinion then why would you show results that are only 50% of your competitors?

    this dd thing makes less sense every day.

  • AnthonySmithAnthonySmith Member, Patron Provider
    # 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, 0.585915 s, 1.8 GB/s
    

    indeed.

  • zonzon Member
    edited November 2012

    I just dont accumulate lebs to get good dd results.. i have 60 sites... and i dont want to put more than 4-5 sites in one vps.. so 17 boxes now...

    i know dd test is just one factor... i give top priority to uptime and overall performance.. my main sites are hosted with inceptionhosting, quickweb, hostigation, buyvm and chicagovps... that doesnt mean others are bad.. some boxes like securedragon and mini vps worked well but because of low mem (128) i was not able to install virtualmin..

    i really had problems with only few... like avante hosting... alienlayer (but now they seems to be better.. but slow...)

  • @zon

    Your websites do writes more than reads?

  • AnthonySmithAnthonySmith Member, Patron Provider
    edited November 2012

    @zon yeah no offence intended, but this bench mark has almost nothing to do with how a site will perform in the vast majority of cases :)

  • @eLohkCalb

    my main sites are photo gallery types.. so mostly read... and we upload pics 1-2 times daily only...

  • @AnthonySmith thats is correct... but with my experience if a dd test is giving very less values like 20 or 30 most of the times then it will affect the site performance... i had some boxes like that and i closed it..

Sign In or Register to comment.