Howdy, Stranger!

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


Hostbench.io - VPS benchmarking tool
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.

Hostbench.io - VPS benchmarking tool

Hostbench.io is automated benchmark tool, which tests server hardware in some ways similar to Serverbear (which, unfortunately, was closed about a week ago)

This benchmark tries to be refined and more meaningful benchmarking tool then Serverbear. Hostbench will also provide the database with all executed benchmarks.

You can test your VPS by getting the link to the script on Hostbench.io by clicking on "benchmark your server" in the navbar. After filling out information about your hosting provider, you will receive a link to the benchmark script, which will look similar to this:

wget -N http://submit.hostbench.io -O hb.sh && sudo bash hb.sh 'YOUR_BENCH_ID'

After the test will be finished, you will resieve a link to your benchmark (Like this).

The benchmark will take about 30 minutes, meanwhile you can read, what exactly will be tested and what are the new things that hostbench.io offers as a VPS benchmark tool.

CPU

CPU benchmark tests the performance by calculating a lot of prime numbers, score depends on the time it took to finish the calculation. This benchmark scales perfectly on CPU's with multiple cores.

Serverbear hadn't had a dedicated CPU benchmark, CPU was tested only as part of UnixBench. Raw CPU performance was not measured by any sort of score, and made comparison between processors in different servers almost impossible.

Storage

Hostbench.io tests storage speed in six different scenarios (4k/64k/512k read/write speed)

The good old Serverbear used dd to test read and write speed, which was basicaly pointless (sequetial speed means almost nothing in real-live use cases).

Read/Write of 4k blocks will tell, how fast the drive handles work with a ton of small files (4k block represents about 40% of the server total disk operations, according to nimblestorage)

64k blocks represent about 30% of total operations,512k block speed is almost identical to sequential speed, and it is relevant for about 15% of the total disk operations.

RAM

Also hostbench.io tests memory bandwidth.

Networking

As we learned through the development, networking is one of the trickiest things to benchmark. The average speed can fluctuate considerably throught the day, the only really solid thing to test was the ping. We tried to make network benchmark very consistant, but in the end is all came down to a compromise, not an ideal solution.

So, Hostbench.io tests the speed between your VPS and several powerful IBM server farms in different parts of the world to measure the maximum achievable speed between a single user and a server. So yes, it is not complete equivalent to the maximum bandwidth of your channel, which you can measure with something like Speedtest.

By the way, during the development we also tried to use speedtest as a tool for measuring download/upload speed, but it didn't turned out so well; the tool was built for average consumer, and when it comes to really high speeds (more than 50 MiB/s or 500MB/s) speedtest results are very inconsistent and unreliable.

The benchmark sourse code is available on GitHub.

We are opened to your feedback! Feel free to leave it here or on hostbench.io twitter.

Comments

  • Looks good. Will do some testing this weekend.

  • MikeAMikeA Member, Patron Provider
    edited August 2016

    "Your connection is not secure"

    Edit: Works in Chrome not Firefox
    Edit 2: Benchmark script doesn't work on OpenVZ CentOS, sudo errors.
    Edit 3: Script doesn't work on Ubuntu 16.04 either, none of the packages are installed/built (fio, bc, sysbench, aria2c errors etc etc.)

    Thanked by 2GCat MikePT
  • first to say: graphics of result looks good, interesting project anyways!

    had a short look at your shell script too.

    I'd suggest not relying on the user providing an ID. what happens if the ID is already taken? haven't seen any check for that esp. before the benchmarks are executed...

    and while you are uninstalling the previous compiled versions of software you download from you git you don't remove packages you are installing via apt before.
    I think you should at least mention in the documentation what is going to be installed, how much disk this may consume space and what probably will be left behind...

    also I don't know if only using softlayer testfiles for the speedtests gives a reliable result... think of different peerings per provider which may result in bad speeds to softlayer but maybe better speeds to other providers...

    just a few first thoughts, will be interested to see how this develops ;-)

  • NeoonNeoon Community Contributor, Veteran

    Server negotiated HTTP/2 with blacklisted suite

  • Cant open the site. "Error code: NS_ERROR_NET_INADEQUATE_SECURITY"

  • Awmusic12635Awmusic12635 Member, Host Rep

    So it seems to have ran fine, but it fails to submit it:

    Full output of script: https://paste.ee/p/KQ4da

  • LomandLomand Member
    edited August 2016

    @MikeA said:
    Edit 3: Script doesn't work on Ubuntu 16.04 either, none of the packages are installed/built (fio, bc, sysbench, aria2c errors etc etc.)

    That is quite strange, just retested the script both on fresh Ubuntu 16.04 x64 and Ubuntu 16.04 x86, worked like a charm.
    Can you provide some information about the errors you getting, maybe you can show the output of the script via https://paste.ee/?

  • @Awmusic12635 said:
    So it seems to have ran fine, but it fails to submit it:

    It looks like the benchmark fails at getting information about CPU if VPS is using OpenVZ.

    Fix incoming. Thank you for your feedback!
    PS: the rest of the benchmark worked fine, so I can manualy add it to the database.
    You have a copy of the result located at /root/benchmark.results, can you paste it in https://paste.ee/ and give me a link?

  • Just added to the script compatibility for OpenVZ servers and fixed some other small bugs. Thanks everyone for your feedback! I'm really sorry that I'm quite short on time and right now I don't have the time to ansver the rest of your questions, but I promise to message you in next 12 hours.
    Cheers!

  • Getting 502 bad gateway on submit.hostbench.io

  • ZappieZappie Member, Host Rep, LIR

    Should consider adding to the network tests both Seoul, South Korea and Chennai, India IBM POPs

    They were probably added/announced post the project being created

  • All these tests don't know why focus on incoming DOWNLOAD speed while for a typical webserver the outgoing UPLOAD speed is more of a concern.

    perhaps something derived from http://ping.pe/ would be ideal to measure network performance ?

  • Sorry, your OS is not supported, therefore, you cant execute this benchmark. Hostbench is compatible with Ubuntu 16.04, 14.04 and Centos 7.

    Seems that your script should be more compatible for other OS like Debian.

  • Centos 7 is now supported! Debian will be available in few days.

  • My Arubacloud bench 1st in value per dollar leaderboards :D

  • Now benchmark supports Debian 6 and 7.
    Waiting for your feedback!

  • No centos 6.8 ?

  • I've used Hostbench.io several times and I've noticed that after about 24 hours I can no longer view my reports on the website. Is that by design?

  • LomandLomand Member
    edited September 2016

    @atlbyron said:
    I've used Hostbench.io several times and I've noticed that after about 24 hours I can no longer view my reports on the website. Is that by design?

    No, not at all. Two days ago Hostbench changed the way permalinks are generated and from then old links (which were send to you by email) were no longer valid, and that is why you were redirected to the homepage when you tried to open it. You can still find the benchmark in "Resent submissions" section, or just PM me with your old link, I'll send you back the new one.
    Anyway, thanks for your feedback!

  • LomandLomand Member
    edited September 2016

    @mgilang said:
    No centos 6.8 ?

    Recentry support for Centos 7 was added. On the other hand, Centos 6 is a bit more tricky to implimend, so for not this distro is now supported. Ecpect this to change in the near future, though. Centos 6 will be supported in about a week or so.

  • Nice work, @Lomand. I tried it on a couple of Debian 8 servers. It's a very straightforward no-bullshit benchmark tool.

  • skarzskarz Member

    Is this project dead @Lomand?

  • digijdigij Member

    @skarz said:
    Is this project dead @Lomand?

    +1

    I have recently realised that the website seems go be gone. I'm wondering why @Lomand - it has been one of the better benchmarking portals.

Sign In or Register to comment.