Howdy, Stranger!

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


TinyKVM - Quality Dallas/Los Angeles KVM SSD VPS - $15/yearly - 10 Years of Excellence - 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.

TinyKVM - Quality Dallas/Los Angeles KVM SSD VPS - $15/yearly - 10 Years of Excellence

2»

Comments

  • uptimeuptime Member
    edited July 2019

    @vyas11 - just pinging to indicate significant info added in edits above ^ ...

    tl;dr: iperf is where it's at - and tinyKVM can haz decent upload bandwidth too.

  • sanvitsanvit Member

    @uptime I have ordered mine and awaiting. As I wait, would you be kind enough to run these two commands for me please?

    iperf -c cloudflare.com -p 443 -i 1
    iperf -c hnd-jp-ping.vultr.com -p 80 -i 1

    Thanks :)

    Thanked by 1uptime
  • uptimeuptime Member
    edited July 2019

    @sanvit - network is fast my man!

    # iperf -c cloudflare.com -p 443 -i 1
    ------------------------------------------------------------
    Client connecting to cloudflare.com, TCP port 443
    TCP window size: 85.0 KByte (default)
    ------------------------------------------------------------
    [  3] local 199.180.xxx.xxx port 58140 connected with 198.41.215.162 port 443
    [ ID] Interval       Transfer     Bandwidth
    [  3]  0.0- 1.0 sec  92.1 MBytes   773 Mbits/sec
    [  3]  1.0- 2.0 sec   100 MBytes   839 Mbits/sec
    [  3]  2.0- 3.0 sec  91.4 MBytes   767 Mbits/sec
    [  3]  3.0- 4.0 sec  91.9 MBytes   771 Mbits/sec
    [  3]  4.0- 5.0 sec  90.2 MBytes   757 Mbits/sec
    [  3]  5.0- 6.0 sec  86.4 MBytes   725 Mbits/sec
    [  3]  6.0- 7.0 sec  87.4 MBytes   733 Mbits/sec
    ^C
    
    # iperf -c hnd-jp-ping.vultr.com -p 80 -i 1
    ------------------------------------------------------------
    Client connecting to hnd-jp-ping.vultr.com, TCP port 80
    TCP window size: 85.0 KByte (default)
    ------------------------------------------------------------
    [  3] local 199.180.xxx.xxx port 50690 connected with 108.61.201.151 port 80
    [ ID] Interval       Transfer     Bandwidth
    [  3]  0.0- 1.0 sec  6.38 MBytes  53.5 Mbits/sec
    [  3]  1.0- 2.0 sec  16.9 MBytes   142 Mbits/sec
    [  3]  2.0- 3.0 sec  12.5 MBytes   105 Mbits/sec
    [  3]  3.0- 4.0 sec  11.1 MBytes  93.3 Mbits/sec
    [  3]  4.0- 5.0 sec  10.8 MBytes  90.2 Mbits/sec
    [  3]  5.0- 6.0 sec  10.5 MBytes  88.1 Mbits/sec
    [  3]  6.0- 7.0 sec  11.2 MBytes  94.4 Mbits/sec
    ^C
    
    Thanked by 1sanvit
  • sanvitsanvit Member

    Lol 100Mbps to JP.! That's awesome! Thanks :)

    Thanked by 1uptime
  • jsgjsg Member, Resident Benchmarker

    @uptime said:
    ... @jsg may have some friendly advice to share along these lines ...

    ???

  • uptimeuptime Member

    @jsg said:

    @uptime said:
    ... @jsg may have some friendly advice to share along these lines ...

    ???

    How to benchmark network performance without getting null-routed ...?

  • ramnetramnet Member, Host Rep

    @uptime said:

    How to benchmark network performance without getting null-routed ...?

    As a general rule: bigger packets are better than smaller packets, and don't sustain the transfer very long.

    Thanked by 2uptime nocom
  • jsgjsg Member, Resident Benchmarker
    edited July 2019

    @uptime said:
    How to benchmark network performance without getting null-routed ...?

    I did quite a bit of research on that for my vpsbench. One factor is obviously to not create lots and lots of traffic.The good news I found is that "downloading" 100 MB is enough (actually beyond 64 MB virtually nothing changes). The other factor is to do the testing in "slices" which actually corresponds nicely to the way a network works anyway. But it comes with a "but": you need high resolution timing (microseconds). Having that available (which most benchmarks don't) you do many small network reads with small pauses in between (a couple of milliseconds) but you carefully measure and add up only the working time. The third factor is obviously to not take name lookup into the timing measurement (which unfortunately most benchmark scripts do and which can badly bend the results).

    Note: This is only low precision in local area networks, so use that only for testing hosts that are a couple of ten milliseconds away (which usually is the case in our benchmarks here).

    Note 2: This is still not a high quality test (for diverse reasons, one of which is that we target http) but it's a lot better than the usual crude script toys and btw also allows to learn more about a target than merely "network speed". I actually use my benchmark (in a special mode showing the "guts" of a connection and transfer) as a simple quick internet analyzer. One point I found very interesting is what might be called "connection establishment and settling". One can see and learn a lot by closely observing how the speed rises which is due to connection establishment and settling of the systems along the route. In a primitive quick and dirty benchmark this can't be seen and is not considered important but fact is that no matter how fast the target server (the box you want to test) is the systems in between the target and yourself almost always draw down performance quite a lot until all of them are settled.

    Thanked by 3uptime vyas11 ITLabs
  • uptimeuptime Member

    @jsg thanks for explaining all that so clearly - definitely raises the bar quite a bit for low-end speedtests!

    Thanked by 1ITLabs
  • KeilKeil Member

    Good stuff. Installed alpine linux and ditched hardened kernel. Hasn't rebooted yet.

    $ uprecords
    # Uptime | System Boot up
    ------------------------------------------+---------------------------------------------------
    -> 1 555 days, 02:52:26 | Linux 4.9.65 Thu Jan 18 01:40:19 2018
    2 7 days, 11:21:54 | Linux 4.9.65-1-virtharded Fri Jan 5 02:40:41 2018

    Anybody requested additional bandwidth? Like, +300mb/yr. I'm afraid the price won't be so tiny.

    Thanked by 1uptime
  • jsgjsg Member, Resident Benchmarker

    FWIW I just had a look at that speedtest-cli thingy and I am between shocked and appalled. It's fat beast with a high degree of boilerplate and gadgets and its time measuring is pretty much worthless. Oh and btw offering benchmarks against a httpS target is mindless due to more than just the factor of session establishment being expensive.

    Thanked by 1uptime
  • uptimeuptime Member
    edited July 2019

    jsg said: between shocked and appalled

    This is generally about the extent of my expressive range as well.

    With occasional interludes of bemusement, and rare episodes of maniacal laughter.

    Thanks for the insight wrt speedtest-cli and the finer points of benchmarking in general - much appreciated!

    Thanked by 1ITLabs
  • ITLabsITLabs Member

    @uptime said:
    With occasional interludes of bemusement, and rare episodes of maniacal laughter.

    Thanked by 2vyas11 uptime
  • vyas11vyas11 Member
    edited July 2019

    @ITLabs said:

    @uptime said:
    With occasional interludes of bemusement, and rare episodes of maniacal laughter.

    That and some language lessons from the master himself... Minute 2:15 onwards

    Thanked by 2uptime ITLabs
  • uptimeuptime Member

    sometimes sense of humor may be the only good sense I have left, really ...

    Thanked by 1ITLabs
Sign In or Register to comment.