Post4VPS Forum | Free VPS Provider
How reliable is Ookla Speedtest for comparing speed of two VPSs? - Printable Version

+- Post4VPS Forum | Free VPS Provider (https://post4vps.com)
+-- Forum: VPS Discussion (https://post4vps.com/Forum-VPS-Discussion)
+--- Forum: VPS Reviews (https://post4vps.com/Forum-VPS-Reviews)
+--- Thread: How reliable is Ookla Speedtest for comparing speed of two VPSs? (/Thread-How-reliable-is-Ookla-Speedtest-for-comparing-speed-of-two-VPSs)

How reliable is Ookla Speedtest for comparing speed of two VPSs? - deanhills - 03-17-2020

How reliable is Ookla Speedtest for comparing the speed of two VPSs?  I'm trying to compare my Phoenix VPS 9 speed with that of Dallas VPS 9.

After the upgrade of my VPS 9 Phoenix last November from OpenVZ 6 to KVM, my IP was changed.  Then when I tested the speed of the VPS it came up with similar results as today's test, like with a HUGE Ping and saying location is Frankfurt, whereas the location is really US in Phoenix.  So how reliable is that really?  Like with that huge Ping I should feel a huge latency issue with Phoenix, but I don't.  My VPS 9 Phoenix performs just as seamless as Dallas, although Dallas feels faster in speed when I work with it (disk speed shows much greater speed as well - refer Benchmark test results of the two VPSs below).  Although my ISP does work faster with Dallas.

Below is the Ookla Speedtest that was taken of VPS 9 Phoenix:

[Image: 8jMUgDX.png]

And this is the Speedtest of VPS 9 Dallas:

[Image: CD01t07.png]

I've done a quick Benchmark test to compare the two (credit to @"Hidden Refuge"):

Code:
wget --no-check-certificate https://raw.githubusercontent.com/hidden-refuge/bench-sh-2/master/bench.sh && bash bench.sh -b && rm bench.sh

VPS 9 Phoenix

System Info
-----------
Processor       : Intel® Xeon® CPU E5-2620 v3 @ 2.40GHz
CPU Cores       : 2
Frequency       : 2399.996 MHz
Memory          : 7870 MB
Swap            : buffers/cache: MB
Uptime          : 86 days, 11:25,


Speedtest (IPv4 only)
---------------------
Your public IPv4 is 50.2.xx.xx

Location                Provider        Speed
CDN                     Cachefly        66.1MB/s

Lenoir, NC, US          Dacentec        28.5MB/s
Dallas, TX, US          Softlayer       43.1MB/s
Seattle, WA, US         Softlayer       34.7MB/s
San Jose, CA, US        Softlayer       52.8MB/s
Washington, DC, US      Leaseweb        30.1MB/s

Tokyo, Japan            Linode          15.7MB/s
Singapore               Softlayer       7.65MB/s
Taiwan                  Hinet           7.25MB/s

Rotterdam, Netherlands  id3.net         14.7MB/s
Haarlem, Netherlands    Leaseweb        10.5MB/s


Disk Speed
----------
I/O (1st run)   : 129 MB/s
I/O (2nd run)   : 108 MB/s
I/O (3rd run)   : 87.6 MB/s
Average I/O     : 108 MB/s or .10 GB/s

Dallas

System Info
-----------
Processor       : Intel® Xeon® CPU E5-2670 v2 @ 2.50GHz
CPU Cores       : 2
Frequency       : 2499.998 MHz
Memory          : 7870 MB
Swap            : buffers/cache: MB
Uptime          : 2:48,


Speedtest (IPv4 only)
---------------------
Your public IPv4 is 23.94.xx.xx

Location                Provider        Speed
CDN                     Cachefly        81.3MB/s

Lenoir, NC, US          Dacentec        53.6MB/s
Dallas, TX, US          Softlayer       57.4MB/s
Seattle, WA, US         Softlayer       30.2MB/s
San Jose, CA, US        Softlayer       33.9MB/s
Washington, DC, US      Leaseweb        25.4MB/s

Tokyo, Japan            Linode          10.2MB/s
Singapore               Softlayer       6.91MB/s
Taiwan                  Hinet           6.20MB/s

Rotterdam, Netherlands  id3.net         27.2MB/s
Haarlem, Netherlands    Leaseweb        80.6MB/s


Disk Speed
----------
I/O (1st run)   : 522 MB/s
I/O (2nd run)   : 473 MB/s
I/O (3rd run)   : 612 MB/s
Average I/O     : 535 MB/s or .53 GB/s

RE: How reliable is Ookla Speedtest for comparing speed of two VPSs? - AmirGT - 03-17-2020

Hello. As you said the VPS is not in Frankfurt but Ookla detects that you are in Frankfurt (bad IP geolocation) so it tests to German servers which is why you get high ping, I suggest you to know where exactly is the VPS located (which city) then search for this city at speedtestserver.com which is Phoenix in your case and you'll probably find some servers in there, specify the server in the speedtest command and check the results now. Also I suggest you to use the official Speedtest CLI which gives much more better results than the Python one, click HERE.

RE: How reliable is Ookla Speedtest for comparing speed of two VPSs? - humanpuff69 - 03-17-2020

on high speed internet it is great . but on extremely low speed internet speedtest.net sometime doesnt load at all or show an error . for comparing a speed of two vps you can target the server to most of the visitor or your location to get the idea for how much speed the user will get

RE: How reliable is Ookla Speedtest for comparing speed of two VPSs? - fChk - 03-17-2020

@deanhills

The Dallas VPS seems superior, regardless of the internet connection speed; Just look at its disk I/O stats (that's the SSD norm). I'm not sure for your use-cases, but for mine disk I/O tend to be a constant bottleneck.

Thus if I were you, I would switch to it without any hesitation :-)

Did you check the SELinux issue on it ?

RE: How reliable is Ookla Speedtest for comparing speed of two VPSs? - deanhills - 03-18-2020

(03-17-2020, 09:24 PM)fChk Wrote: @deanhills

Thus if I were you, I would switch to it without any hesitation :-)
Absolutely what has been going through my mind.  We're using it for testing Ubuntu right now for installation on a VPS Holder's VPS, but with the next Giveaway in April I intend to apply for VPS 9 Dallas.  I really love this VPS. Cool

Anyway, thanks very much for the location tool you sent.  This is much appreciated.  I was able to locate the IP in Tempe, Arizona.

@AmirGT Thank you for the comment.  That makes absolute sense to me.
OK, with all of your help I was able to get an updated speed test for VPS 9 Phoenix.  

@fChk provided me with a tool to find the real location of the IP for VPS 9 Phoenix which is Tempe, Arizona (not Germany).

As @AmirGT mentioned one can search for the closest speedtest server location at https://www.speedtestserver.com/

The closest available speedtest server is:
Nogales, Arizona Server ID: 16960

When I Googled for the script this is the Website I found that showed me how to do the speedtest with Server ID:
https://www.cyberciti.biz/faq/install-speedtest-cli-on-centos-redhat-fedoa-scientific-to-measure-internetspeed/

Code:
./speedtest-cli --server 16960 --share

And this was the result.  Looks a bit more realistic now:

[Image: 9144003813.png]

Dallas definitely compares better so am planning to apply for VPS 9 Dallas during the next Giveaway.  I must say VPS 9 Phoenix, even though it is still an excellent VPS deteriorated significantly in speed after the upgrade.  This is the last speed I took just before the upgrade to KVM and change of IP for the VPS:

[Image: 0Sm5MwN.png]

RE: How reliable is Ookla Speedtest for comparing speed of two VPSs? - AmirGT - 03-18-2020

(03-18-2020, 01:31 AM)deanhills Wrote: And this was the result.  Looks a bit more realistic now:

[Image: 9144003813.png]

That still doesn't seem to be a close location.
A month ago I got a VPS and I wanted to run speedtest so I downloaded the Python speedtest script and ran it without specifying a server and this was the result

[Image: 9074472302.png]

Good, right? I shared it with my friend and he told me to use the official speedtest CLI because it gives better results. I installed it and, specified a close server (UPC Romania) then ran speedtest and that was the result

[Image: c9e268f8-bb16-4821-8108-548ec26cf67b.png]

I even tried the same server with speedtest Python script and the official speedtest CLI, the official speedtest CLI was much faster. Did you know now why I asked you to use the official speedtest CLI and specify a close server?

Also about locating the IP you can use check-host.net it works very great and it also supports IPv6

RE: How reliable is Ookla Speedtest for comparing speed of two VPSs? - deanhills - 03-19-2020

(03-18-2020, 05:16 PM)AmirGT Wrote: That still doesn't seem to be a close location.
A month ago I got a VPS and I wanted to run speedtest so I downloaded the Python speedtest script and ran it without specifying a server and this was the result

[Image: 9074472302.png]

Good, right? I shared it with my friend and he told me to use the official speedtest CLI because it gives better results. I installed it and, specified a close server (UPC Romania) then ran speedtest and that was the result

[Image: c9e268f8-bb16-4821-8108-548ec26cf67b.png]

I even tried the same server with speedtest Python script and the official speedtest CLI, the official speedtest CLI was much faster. Did you know now why I asked you to use the official speedtest CLI and specify a close server?

Also about locating the IP you can use check-host.net it works very great and it also supports IPv6
@AmirGT Actually that is exactly what I thought I did if you look a bit closer at my previous post. I specified the "closest" server to the location of the VPS. And I added it with the official speedtest CLI. I've been using the official speedtest CLI with all my VPSs. Only issue I had was when the IP I got that was used for Tempe, Arizona looked like a German IP. You then helped me in the right direction. I doubt I'm going to get a better result, as when I checked the available servers in Arizona for the speed test, I could only find one in Nogales. But I can always check again later. I was hoping for one in Phoenix. Do you have an ID for a Server in Phoenix, Arizona?

RE: How reliable is Ookla Speedtest for comparing speed of two VPSs? - AmirGT - 03-19-2020

(03-19-2020, 12:23 AM)deanhills Wrote: @AmirGT  Actually that is exactly what I thought I did if you look a bit closer at my previous post.  I specified the "closest" server to the location of the VPS. And I added it with the official speedtest CLI.  I've been using the official speedtest CLI with all my VPSs.  Only issue I had was when the IP I got that was used for Tempe, Arizona looked like a German IP.  You then helped me in the right direction.  I doubt I'm going to get a better result, as when I checked the available servers in Arizona for the speed test, I could only find one in Nogales.  But I can always check again later.  I was hoping for one in Phoenix.  Do you have an ID for a Server in Phoenix, Arizona?

If speedtestserver.com didn't list any then probably there are no servers in there but you could always check the map for closer cities, if that didn't help then I suggest you to use iperf, it doesn't have result links like speedtest but it's more accurate and preferred for servers, just specify the server and go.

RE: How reliable is Ookla Speedtest for comparing speed of two VPSs? - humanpuff69 - 03-21-2020

(03-18-2020, 05:16 PM)AmirGT Wrote: That still doesn't seem to be a close location.
A month ago I got a VPS and I wanted to run speedtest so I downloaded the Python speedtest script and ran it without specifying a server and this was the result

[Image: 9074472302.png]

Good, right? I shared it with my friend and he told me to use the official speedtest CLI because it gives better results. I installed it and, specified a close server (UPC Romania) then ran speedtest and that was the result

[Image: c9e268f8-bb16-4821-8108-548ec26cf67b.png]

I even tried the same server with speedtest Python script and the official speedtest CLI, the official speedtest CLI was much faster. Did you know now why I asked you to use the official speedtest CLI and specify a close server?

Also about locating the IP you can use check-host.net it works very great and it also supports IPv6

the host is the same . the only differences is the ping . maybe at the time of the first one the network is less used at the time
or the host have 2 host on the same location but with the same name

if you want to select the server what i do is install a gui and a browser . but it isnt reccomended on low end vps

RE: How reliable is Ookla Speedtest for comparing speed of two VPSs? - fChk - 04-07-2020

(03-18-2020, 05:16 PM)AmirGT Wrote: That still doesn't seem to be a close location.

(...)

Ok!... Here we go:

> This is what it looks like when speedtest is left to its own logic (ie testing against a Frankfurt-based server):
Code:
user@kvm-Post2Host-Phoenix:~/Lab$ ./speedtest
  Speedtest by Ookla

    Server: 31173 Services AB - Frankfurt (id = 23095)
       ISP: Eonix Corporation
   Latency:   145.83 ms   (1.14 ms jitter)
  Download:   587.04 Mbps (data used: 876.5 MB)                              
    Upload:   413.44 Mbps (data used: 651.2 MB)                              
Packet Loss:     0.0%

> This is what it looks like when speedtest is directed towards a series of servers based in Phoenix Arizona:
Code:
user@kvm-Post2Host-Phoenix:~/Lab$ ./speedtest -s 30226
  Speedtest by Ookla

    Server: Cytranet - Phoenix, AZ (id = 30226)
       ISP: Eonix Corporation
   Latency:     2.29 ms   (0.46 ms jitter)
  Download:   853.87 Mbps (data used: 925.8 MB)                              
    Upload:   914.24 Mbps (data used: 1.1 GB)                              
Packet Loss:     0.0%

user@kvm-Post2Host-Phoenix:~/Lab$ ./speedtest -s 28800
  Speedtest by Ookla

    Server: PhoenixNAP Global IT Services - Phoenix, AZ (id = 28800)
       ISP: Eonix Corporation
   Latency:     0.59 ms   (0.12 ms jitter)
  Download:   579.57 Mbps (data used: 779.8 MB)                              
    Upload:   553.91 Mbps (data used: 1.0 GB)                              
Packet Loss:     0.0%

user@kvm-Post2Host-Phoenix:~/Lab$ ./speedtest -s 30226
  Speedtest by Ookla

    Server: Cytranet - Phoenix, AZ (id = 30226)
       ISP: Eonix Corporation
   Latency:     2.06 ms   (0.06 ms jitter)
  Download:   886.08 Mbps (data used: 1.1 GB)                              
    Upload:   907.28 Mbps (data used: 1.1 GB)                              
Packet Loss:     0.0%

user@kvm-Post2Host-Phoenix:~/Lab$ ./speedtest -s 28800
  Speedtest by Ookla

    Server: PhoenixNAP Global IT Services - Phoenix, AZ (id = 28800)
       ISP: Eonix Corporation
   Latency:     0.63 ms   (0.19 ms jitter)
  Download:   596.34 Mbps (data used: 741.2 MB)                              
    Upload:   471.95 Mbps (data used: 848.1 MB)                              
Packet Loss:     0.0%

user@kvm-Post2Host-Phoenix:~/Lab$ ./speedtest -s 29074
  Speedtest by Ookla

    Server: IOFLOOD.com Dedicated Servers - Phoenix, AZ (id = 29074)
       ISP: Eonix Corporation
   Latency:     0.44 ms   (0.63 ms jitter)
  Download:   938.35 Mbps (data used: 422.3 MB)                              
    Upload:   712.64 Mbps (data used: 716.4 MB)                              
Packet Loss:     0.0%

user@kvm-Post2Host-Phoenix:~/Lab$ ./speedtest -s 10394
  Speedtest by Ookla

    Server: Speedtest.net - Phoenix, AZ (id = 10394)
       ISP: Eonix Corporation
   Latency:     1.00 ms   (0.51 ms jitter)
  Download:   617.69 Mbps (data used: 707.8 MB)                              
    Upload:   697.16 Mbps (data used: 678.6 MB)                              
Packet Loss:     0.0%

user@kvm-Post2Host-Phoenix:~/Lab$ ./speedtest -s 15783
  Speedtest by Ookla

    Server: Sprint - Phoenix, AZ (id = 15783)
       ISP: Eonix Corporation
   Latency:    44.60 ms   (0.53 ms jitter)
  Download:   808.76 Mbps (data used: 917.0 MB)                              
    Upload:   898.55 Mbps (data used: 1.2 GB)                              
Packet Loss:     0.0%

user@kvm-Post2Host-Phoenix:~/Lab$ ./speedtest -s 12219
  Speedtest by Ookla

    Server: QTS Data Centers - Phoenix, AZ (id = 12219)
       ISP: Eonix Corporation
   Latency:     1.11 ms   (2.58 ms jitter)
  Download:   524.51 Mbps (data used: 772.8 MB)                              
    Upload:   473.03 Mbps (data used: 826.7 MB)                              
Packet Loss:     0.0%

user@kvm-Post2Host-Phoenix:~/Lab$ ./speedtest -s 16331
  Speedtest by Ookla

    Server: 1GServers.com - Phoenix, AZ (id = 16331)
       ISP: Eonix Corporation
   Latency:     0.75 ms   (0.44 ms jitter)
  Download:   541.28 Mbps (data used: 608.3 MB)                              
    Upload:   565.78 Mbps (data used: 585.2 MB)                              
Packet Loss:     0.0%

user@kvm-Post2Host-Phoenix:~/Lab$ ./speedtest -s 17614
  Speedtest by Ookla

    Server: CyberLynk Network - Phoenix, AZ (id = 17614)
       ISP: Eonix Corporation
   Latency:     0.42 ms   (0.47 ms jitter)
  Download:   613.14 Mbps (data used: 919.2 MB)                              
    Upload:   852.55 Mbps (data used: 554.5 MB)                              
Packet Loss:     0.0%

user@kvm-Post2Host-Phoenix:~/Lab$ ./speedtest -s 30415
  Speedtest by Ookla

    Server: Janus Networks - Phoenix, AZ (id = 30415)
       ISP: Eonix Corporation
   Latency:     2.07 ms   (9.03 ms jitter)
  Download:   638.00 Mbps (data used: 507.7 MB)                              
    Upload:   727.33 Mbps (data used: 1.3 GB)                              
Packet Loss:     0.0%

user@kvm-Post2Host-Phoenix:~/Lab$ ./speedtest -s 16613
  Speedtest by Ookla

    Server: Cox - Phoenix - Phoenix, AZ (id = 16613)
       ISP: Eonix Corporation
   Latency:    11.78 ms   (0.08 ms jitter)
  Download:   917.80 Mbps (data used: 1.1 GB)                              
    Upload:   315.74 Mbps (data used: 501.9 MB)                              
Packet Loss: Not available.

user@kvm-Post2Host-Phoenix:~/Lab$ ./speedtest -s 8695
  Speedtest by Ookla

    Server: Sparklight - Phoenix, AZ (id = 8695)
       ISP: Eonix Corporation
   Latency:     0.93 ms   (0.02 ms jitter)
  Download:   927.53 Mbps (data used: 569.0 MB)                              
    Upload:   747.36 Mbps (data used: 818.2 MB)                              
Packet Loss:     0.0%

I would say that Phoenix-based VPS-9' connection speed is top-notch!.. It's the decaying US infrastructure and/or the network traffic that makes it look not that good sometimes!