Howdy, Stranger!

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


Building a Serverbear alternative - Page 2
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.

Building a Serverbear alternative

2»

Comments

  • vfusevfuse Member, Host Rep

    @flex said:
    Your service looks pretty nice.
    Please add more providers in the selection.

    soyoustart and even some german providers would be great

    I'm pretty sure that the selection field is a create field as well so if it's not in the list it will be created.

  • flexflex Member

    @vfuse said:

    @flex said:
    Your service looks pretty nice.
    Please add more providers in the selection.

    soyoustart and even some german providers would be great

    I'm pretty sure that the selection field is a create field as well so if it's not in the list it will be created.

    you're right, sorry for this ;)

  • AntonAnton Member

    @flex said:
    you're right, sorry for this ;)

    Thanks for the feedback and sorry that this feature is not obvious.

    I've put "or type in" in the placeholder thinking that it would be enough but turns out I was wrong. I'll think how I can clear this up. Any suggestions?

  • flexflex Member

    i think the or type in is clear enough

  • Anton said: Or may be I should pick a few sets of servers and let users decide if they want only servers in their region or across all continents.

    Why not both? :D

    Yes, for websites, a server close to the audience is perfect, however, I can see other uses for servers being in different locations (they aren't all publicly visible) - connectivity between them could be important. For comparison, Serverbear has Cachefly (basically a speed test to the closest Speedtest location) and other locations around the world.

    Anton said: Closest servers are found using speedtest.net API and I haven't seen it fail so far.

    I've had it fail before (my IP was recently bought and geo takes time to change). It was a server in Texas, but Speedtest thought it was in Florida.

    Anton said: What if I hide raw output by default and have a link there that would unfold these blocks?

    Ooh, perfect.

  • AntonAnton Member

    @Silvenga
    I see your point. I'll figure out a way to implement that.
    Thanks

  • FredQcFredQc Member

    Here is mine:

    Test results for LV E3-KVM 4GB at BuyVM 
    
    Server specs: 
    Intel(R) Xeon(R) CPU E3-1270 v3 @ 3.50GHz 
    3.7 GB  RAM / 80 GB disk space 
    Centos 7.2.1511 Core
    Las Vegas, United States
    
    Benchmark summary:
    UnixBench - 2091.2
    Disk Read - 472 MB/s
    Disk Write - 1478 MB/s
    Bandwidth - 90.43 MB/s
    
    More: https://serverscope.io/trials/R0k
    

    Only problem was that I received the email results with the good IP but... in the SPAM folder. Other than this, good work, love the interface ;)

  • AntonAnton Member
    edited June 2016

    @FredQc said:
    Only problem was that I received the email results with the good IP but... in the SPAM folder.

    Yeah, I've fixed the IP detection. Will look into delivery issues to your email host (outlook.com right?).

    Thanks for the kind words!

  • CPU detection failed
    https://serverscope.io/trials/3Am#system

    cat /proc/cpuinfo
    vendor_id       : IBM/S390
    # processors    : 2
    bogomips per cpu: 20325.00
    features        : esan3 zarch stfle msa ldisp eimm dfp etf3eh highgprs vx
    cache0          : level=1 type=Data scope=Private size=128K line_size=256 associativity=8
    cache1          : level=1 type=Instruction scope=Private size=96K line_size=256 associativity=6
    cache2          : level=2 type=Data scope=Private size=2048K line_size=256 associativity=8
    cache3          : level=2 type=Instruction scope=Private size=2048K line_size=256 associativity=8
    cache4          : level=3 type=Unified scope=Shared size=65536K line_size=256 associativity=16
    cache5          : level=4 type=Unified scope=Shared size=491520K line_size=256 associativity=30
    processor 0: version = FF,  identification = 036A77,  machine = 2964
    processor 1: version = FF,  identification = 036A77,  machine = 2964  
    

    https://serverscope.io/trials/vA7

    cat /proc/cpuinfo
    processor       : 0
    Features        : fp asimd aes pmull sha1 sha2 crc32
    CPU implementer : 0x43
    CPU architecture: 8
    CPU variant     : 0x0
    CPU part        : 0x0a1
    CPU revision    : 0
    
    processor       : 1
    Features        : fp asimd aes pmull sha1 sha2 crc32
    CPU implementer : 0x43
    CPU architecture: 8
    CPU variant     : 0x0
    CPU part        : 0x0a1
    CPU revision    : 0  
    
  • Awmusic12635Awmusic12635 Member, Host Rep

    I would recommend getting rid of the speedtest.net test to instead use more data center speed test files.

    The speedtest.net tests work horribly on servers. Geolocation is most always wrong, the server you are testing from often has more bandwidth available than the speed test server (which leads to incorrect reports) and it is very inconsistent.

  • yomeroyomero Member

    What about geekbench instead of unixbench?

    Thanked by 1tr1cky
  • NomadNomad Member

    Hmm, is this disk read supposed to be this low or my configuration is bad? :S

    https://serverscope.io/trials/eK4#io

  • @Nomad
    Ok for sata hdd.

  • labraxlabrax Member
    edited June 2016

    here's mine from liteserver cheap 1y deals

    Test results for OVZ-100HDD-128 at LiteServer 
    
    Server specs: 
    Intel(R) Xeon(R) CPU E3-1270 v3 @ 3.50GHz 
    128 MB  RAM / 105 GB disk space 
    Ubuntu 14.04 Trusty
    , Netherlands
    
    Benchmark summary:
    UnixBench - 1630.7
    Disk Read - 745 MB/s
    Disk Write - 631 MB/s
    Bandwidth - 436.99 MB/s
    
    More: https://serverscope.io/trials/ZKm
    
    Thanked by 2LiteServer vimalware
  • SaragoldfarbSaragoldfarb Member
    edited June 2016

    @Anton said:

    @Saragoldfarb said:
    Finished the bench. No link returned though.

    Ugh, I'm sorry about that. Could you share provider / plan and OS? I would really like to troubleshoot that. Thank you!

    It was on a Debian install but on a random machine I was logged in already when I noticed your post. Probably a dedi @ online. I'll give it another try on a couple of different installs and report back.

  • tr1ckytr1cky Member

    I wouldn't use unixbench. Not only are the results flawed on some virtualisations (e.g. Xen), it also takes a lot of time.

    A geekbench implementation would be a lot better.

  • AntonAnton Member

    @Strikerr said:
    CPU detection failed

    Thanks for the report! Will add to my list :)

    @Awmusic12635 said:
    I would recommend getting rid of the speedtest.net test to instead use more data center speed test files.

    In that case wouldn't we be testing download to the servers instead of speed of outbound traffic? My assumption is that the vast majority of web servers are being downloaded data from rather than uploaded data to. I've been given a few ideas how to work around failing geo location already. Looking to implement that in the next few weeks.

    Thanks!

    @yomero said:
    What about geekbench instead of unixbench?
    @tr1cky said:
    I wouldn't use unixbench.

    I was thinking about adding geekbench to the kit (since you can pick and choose which benchmarks to run). Not a big fan of UnixBench myself but a lot of people use it.

    Thanked by 1yomero
  • Awmusic12635Awmusic12635 Member, Host Rep

    Anton said: In that case wouldn't we be testing download to the servers instead of speed of outbound traffic? My assumption is that the vast majority of web servers are being downloaded data from rather than uploaded data to. I've been given a few ideas how to work around failing geo location already. Looking to implement that in the next few weeks.

    Well the speedtest website does both upload and download tests when a user runs it. Most every time a customer comes to me saying they are not getting their advertised network speed it is because they are using the speedtest.net cli tool. It is wrong too much and just leads to clients getting incorrect results.

    I always end up telling them that the test is not suitable to test servers. If you want your benchmark to replace serverbear, I would prefer if I could actually recommend it and not tell clients to ignore the network results.

  • SpoofySpoofy Member

    Speedtest sux - use iperf instead.
    Also fio options are incorrect - IOPS results are totally wrong.

  • yomeroyomero Member

    @Spoofy said:
    Speedtest sux - use iperf instead.
    Also fio options are incorrect - IOPS results are totally wrong.

    Iperf needs another server. And I think nobody wants to expose one to the volumes of traffic it needs.

  • AntonAnton Member

    @Awmusic12635 said:
    Well the speedtest website does both upload and download tests when a user runs it.

    I see what you're saying and a I agree that speedtest is not the best solution.

    I've opted out to only test upload (so, how fast the server can serve files).
    I might add download tests later and since you're running a hosting company, could you please share how much inbound traffic you get in comparison to outbound (my assumption would be 10-20%)?

    a customer comes to me saying they are not getting their advertised network speed it is because they are using the speedtest.net cli tool

    Although I see your point, I'm not sure if the tool is to blame here. The issue is most likely that some of your customers don't understand how the whole network thing works. I know some people that constantly get upset that they don't get their full speed at home while downloading from some servers located in the middle of nowhere across the ocean.

    I agree that this can be an issue. What would you suggest we do?

    @Spoofy said:
    Speedtest sux - use iperf instead.

    I'm planning to look more into iperf in later. My initial concern was that I couldn't quickly find public servers available. For example, this page https://iperf.fr/iperf-servers.php lists only two servers in North America both located in California. I would appreciate if you know where to get more.

    Also fio options are incorrect - IOPS results are totally wrong.

    Total file size for the test can be too small on the servers that don't have much RAM and that's on my list to fix.

    Could elaborate why you believe that fio options are incorrect? How would you change them?


    I would like to take a moment and say Thank you for all the feedback this community has already given. I did not anticipate such a warm welcome, you guys are awesome. I'm so glad that I've decided to post here and hope that I can deliver to your expectations.

  • Awmusic12635Awmusic12635 Member, Host Rep

    Anton said: Although I see your point, I'm not sure if the tool is to blame here. The issue is most likely that some of your customers don't understand how the whole network thing works. I know some people that constantly get upset that they don't get their full speed at home while downloading from some servers located in the middle of nowhere across the ocean.

    While I definitely get certain types of things with people downloading across the ocean I am mainly referring to situations like this:

    Most speedtest.net servers seem to have 1Gbps max connection to them. Speedtest.net is designed to test residential connections so most of the time this is fine.

    My nodes each have a 10Gbps connection, 10x that of the server they are testing against. If you combine that with the bandwidth being used by others testing against the speed test server the results are usually far far off, even when in close proximity.

  • AntonAnton Member

    @Awmusic12635 I agree. What would you suggest as an alternative?

    And what about inbound / outbound bandwidth ratio, would you be able to share that info?

  • there is no search

  • Any updates? In august serverbear closed, so I can only imagine there's more interest in a new, reliable comparison site than ever before...

Sign In or Register to comment.