Disk Performance: Slicehost vs. Linode

By Daniel Miessler on December 5th, 2009: Tagged as Linode | Slicehost | Technology
  • http://twitter.com/gionn Giovanni Toraldo

    You should make some more testing in different day/time: performances are highly depending also on the general load of the cluster and of your neighbors.

  • http://robbyt.net/ rob t

    you should use a real benchmark, like iozone. hdparm isn't very accurate.

  • Pingback: uberVU - social comments

  • http://www.linode.com/ Jed Smith

    We have them in beta, but they shouldn't be relied upon just yet. File a ticket and I'll set you up with them (Newark's full, but file a ticket anyway).

  • Israel

    Hi Daniel, what did you use to get this numbers?

  • Israel

    Hi Daniel, what did you use to get this numbers?

  • http://danielmiessler.com/ Daniel Miessler

    hdparm -tT

  • Israel

    Hi Daniel, What tool do you use to get this numbers?

  • http://arik.baratz.org/ Arik

    I use vpslink.com

    Their smallest machine, 64MB RAM, is enough for my asterisk box and the occasional SSH tunnel.

  • http://www.theuniverses.com/ The Universes

    VPSLink offers excellent resources and performance for the price.
    I was curious and ran hdparm -tT on one of my VPS with them and got:
    # hdparm -tT /dev/sda1

    /dev/sda1:
    Timing cached reads: 20104 MB in 2.00 seconds = 10071.99 MB/sec
    Timing buffered disk reads: 778 MB in 3.00 seconds = 259.17 MB/sec

  • Carlos Paredes

    I have just found this site googling. I just have detected the Slicehost decreased speed, because when I joined (in 2008) it was pretty fast. Now I get about 3-10 mb/sec on disk, and previously it was well over 100 Mb/sec. The CPU speed is also slightly worse. It seems that they don't upgrade the hardware and continue charging the same money for even cheaper resources (a shared network disk instead a native one).

    My current (2010) slicehost performance metrics:

    # dd if=/dev/sda1 of=/dev/null bs=1048576 count=1024
    1024+0 records in
    1024+0 records out
    1073741824 bytes (1.1 GB) copied, 117.446 s, 9.1 MB/s

    $ time pi 1000000 > /dev/null

    real 0m12.218s
    user 0m6.140s
    sys 0m0.050s

    My old (2008) slicehost performance metrics:

    # dd if=/dev/sda1 of=/dev/null bs=1048576 count=1024
    1024+0 records in
    1024+0 records out
    1073741824 bytes (1.1 GB) copied, 10.3052 seconds, 104 MB/s

    $ time pi 1000000 > /dev/null

    real 0m10.876s
    user 0m10.673s
    sys 0m0.204s

    Once my current credit expires (in about 12 months) I'll likely switch to another provider.

  • Carlos Paredes

    I have just found this site googling. I just have detected the Slicehost decreased speed, because when I joined (in 2008) it was pretty fast. Now I get about 3-10 mb/sec on disk, and previously it was well over 100 Mb/sec. The CPU speed is also slightly worse. It seems that they don't upgrade the hardware and continue charging the same money for even cheaper resources (a shared network disk instead a native one).

    My current (2010) slicehost performance metrics:

    # dd if=/dev/sda1 of=/dev/null bs=1048576 count=1024
    1024+0 records in
    1024+0 records out
    1073741824 bytes (1.1 GB) copied, 117.446 s, 9.1 MB/s

    $ time pi 1000000 > /dev/null

    real 0m12.218s
    user 0m6.140s
    sys 0m0.050s

    My old (2008) slicehost performance metrics:

    # dd if=/dev/sda1 of=/dev/null bs=1048576 count=1024
    1024+0 records in
    1024+0 records out
    1073741824 bytes (1.1 GB) copied, 10.3052 seconds, 104 MB/s

    $ time pi 1000000 > /dev/null

    real 0m10.876s
    user 0m10.673s
    sys 0m0.204s

    Once my current credit expires (in about 12 months) I'll likely switch to another provider.

  • Pingback: RSS Tidbits for 5 July 2010 « SKFox.com

  • Patrick

    I decided to try out Linode because Slicehost's backbone (CogentCo) between the EU and US was down (99%+ packet loss) for half a week, and immediately noticed that the Linode was as fast as my Slice had been when I first got it, and that I wasn't just imagining a slowdown. A popular argument in favor of Xen providers is that they can't possibly oversell CPU and RAM, but you have to think about disk and network too.

    I'm satisfied with Linode, but they have a lot of network issues. I had to move one slice away from their Dallas data center completely because of several long outages/maintenances on the backbone per week, but at least I had the option of doing so, and they were quite helpful. There are a few network issues (slowdowns during peak hours, primarily) where I am now (Absecon, NJ data center), but I'm not going back to Slicehost in the foreseeable future.

  • Patrick

    Damn, I'm still calling them “slices”. “I had to move one Linode away”, even :)

  • http://www.facebook.com/JshWright Josh Wright

    The Linode offices are in Absecon (south Jersey), the NJ data center is NAC's Cedar Knolls DC (north Jersey).

  • Patrick

    Ah, thanks for the clarification.

  • http://www.iphone4forum.net iPhone 4 Forum

    I've been a Linode customer for about a year now and have been happy with the service.

  • Steve

    Linode does have backup snapshots now ;)

  • Pingback: Jason Mooberry » Blog Archive » A new home

  • http://www.americanrecordablemedia.com/ Buy Blank CD

    How about the price of these disks? How much is their difference?

  • Pingback: Slicehost vs Linode | ahmedkamel.net


Top

Popular

Information Security / Technology

Politics

Philosophy & Religion

Technology & Science

Culture & Society

Miscellaneous

Arguments

Projects

Collections

Twitter

What I'm Reading

Favorite Books and Essays

Top Blog Categories

Inputs