25 replies to this topic

[thomasrelee]

I've been conducting speed tests using your service for a while now.  I would use various services to check my speed and also participate in the SamKnows Broadband testing - all speed tests used to agree with one another from service to service, with the exception of speedtest.net - which always shows results inflated by PowerBoost. I fully understand the technical aspect involved, including peak hours congestion of the Internet and other issues that can affect measurements.

On September 27, my area completed upgrades with Docsis 3.0 and  was far enough along with the upgrade, they went ahead and bumped the speeds up for the lower Standard and Turbo tiers.  I subscribe to the Turbo Tier, which was 10 mbps, and after the bump in speed, the tier is now rated at 15 mbps.

Since this upgrade, I have never been able to consistently test up to the 15 mbps full speed of my connection against specifically testmy.net.  I have run HTTP tests, FTP tests, Java Tests, Flash Tests, etc - and do not seem to have many problems with testing from any service except this one.  I do not understand where the problem is.

A quick snapshot of what I have here:

- A brand new (six-month old) RCA DCM425 cable modem provided by TWC.
- A Linux firwall/router running ClearOS using an Intel d510mo Atom system I threw together with 2GB RAM and a 500GB HD.
.....- feeding into a TrendNet 1Gbps switch with the following connect to the switch:
..........- Two PC's,
..........- One SamKnows testing device (operating in bridge mode, with nothing connected to it directly)
..........- and a Vonage adapter connected to this switch.
- All connections are using CAT-6

I first realized I was not running QOS on the Linux firewall/router, and had an idea to turn it on and see what would happen.  Once I did this on September 27, I started being able to test to my maximum line rate of 15 mbps against testmy.net.  Though I do not understand why this would have any effect on anything - it should not.  What was weird though, is when I would turn off QOS, the tests would indicated lower speeds, and when I turned it back on, it would indicate full rated speeds.  All of this occurring on September 27 - 29.  Once I started receiving results that were steady and satisfactory, I left QOS on and called it fixed.

A few days passed and I come back to check speeds again - now there is nothing I can do to even come close to 15 mbps.  Again, traceroutes are not showing anything wrong.  I even took and directly connect my PC to the cable modem and the results were still low (6 - 10 mbps for a 15 mbps connection).  These tests were conducted repeatedly night and day Oct 1 and Oct 2

I did not record the PC's connection ID or log in for the direct connected tests, but I am fully aware of the connection ID running my normal  setup.  It is 160470957976.  I have no other problem with any one speed test not showing my connections maximum rated speed - even during peak hours.  

A quick history on my connection...  Up until July 14, 2011, my connection had been running horrible for 10 months.  A combination of a congested cable node in addition to other things the cable company was doing wrong were to blame.  It was finally resolved and everything has been smooth sailing until now.  I first though - great, my cable company has screwed up the connection again, but it is not - and is only with connections to testmy.net I am experiencing problems.  

It is odd and I am curious to know what is going on here.  I love the testmy.net service, but it does not currently love me for some reason.

I suspect it is a peering problem, and not on TW's part...  Every where else I honestly have no issues.  Below is a traceroute I run multiple times during a speed test I performed before hitting POST.  I do actually see several problems in these traceroutes between hops 7 & 8 where hop 8 spikes close to or over 200ms:

203.466 ms  186.419 ms  186.297 ms
53.254 ms  241.074 ms  241.300 ms
236.022 ms  220.118 ms  220.042 ms

Again, these traces were run while I was running the actual speed test. Validation Link:: http://testmy.net/db/6BeiOJf


There is even one trace below that indicates a spike between hop 6 & 7 of:  29.847 ms  172.769 ms  172.683 ms

These traceroutes are run directly from my firewall/router, so they start reading hop one at my cable modem's gateway address and not the IP of the connection...


[user@clearworld1 ~]# tracert 174.120.187.140
traceroute to 174.120.187.140 (174.120.187.140), 30 hops max, 40 byte packets
1  10.241.192.1 (10.241.192.1)  8.824 ms  8.688 ms  8.587 ms
2  gi15-0-0-3101.gnboncsg-rtr2.triad.rr.com (24.28.225.189)  11.157 ms  11.543 ms  11.569 ms
3  24.93.64.76 (24.93.64.76)  14.816 ms  14.863 ms  14.875 ms
4  107.14.19.18 (107.14.19.18)  19.183 ms  19.232 ms  19.236 ms
5  107.14.19.11 (107.14.19.11)  44.031 ms  44.043 ms  44.028 ms
6  TenGigabitEthernet9-2.ar1.ATL2.gblx.net (64.212.108.69)  19.728 ms  18.663 ms  18.621 ms
7  The-Planet-Dallas.TenGigabitEthernet6-2.ar5.DAL2.gblx.net (67.17.168.94)  44.710 ms  44.831 ms  42.097 ms
8  te2-4.dsr01.dllstx3.networklayer.com (70.87.255.38)  42.074 ms  49.523 ms  49.505 ms
9  * * *
10  a.ff.5746.static.theplanet.com (70.87.255.10)  44.792 ms  44.842 ms  66.383 ms
11  8c.bb.78ae.static.theplanet.com (174.120.187.140)  65.506 ms  42.885 ms  42.871 ms
[user@clearworld1 ~]# tracert 174.120.187.140
traceroute to 174.120.187.140 (174.120.187.140), 30 hops max, 40 byte packets
1  10.241.192.1 (10.241.192.1)  9.775 ms  9.750 ms  9.758 ms
2  gi15-0-0-3101.gnboncsg-rtr2.triad.rr.com (24.28.225.189)  10.933 ms  10.965 ms  10.969 ms
3  24.93.64.76 (24.93.64.76)  13.404 ms  13.441 ms  13.435 ms
4  107.14.19.18 (107.14.19.18)  18.707 ms  18.755 ms  18.760 ms
5  107.14.19.11 (107.14.19.11)  18.708 ms  18.696 ms  18.739 ms
6  TenGigabitEthernet9-2.ar1.ATL2.gblx.net (64.212.108.69)  57.626 ms  31.138 ms  31.001 ms
7  The-Planet-Dallas.TenGigabitEthernet6-2.ar5.DAL2.gblx.net (67.17.168.94)  43.587 ms  47.134 ms  47.119 ms
8  te2-4.dsr01.dllstx3.networklayer.com (70.87.255.38)  47.062 ms  44.470 ms  44.417 ms
9  * * *
10  a.ff.5746.static.theplanet.com (70.87.255.10)  48.150 ms  47.720 ms  47.706 ms
11  8c.bb.78ae.static.theplanet.com (174.120.187.140)  46.562 ms  43.134 ms  43.180 ms
[user@clearworld1 ~]# tracert 174.120.187.140
traceroute to 174.120.187.140 (174.120.187.140), 30 hops max, 40 byte packets
1  10.241.192.1 (10.241.192.1)  9.106 ms  9.075 ms  9.067 ms
2  gi15-0-0-3101.gnboncsg-rtr2.triad.rr.com (24.28.225.189)  10.701 ms  10.753 ms  10.758 ms
3  24.93.64.76 (24.93.64.76)  12.923 ms  12.973 ms  12.960 ms
4  107.14.19.18 (107.14.19.18)  18.739 ms  18.793 ms  18.779 ms
5  107.14.19.11 (107.14.19.11)  18.761 ms  18.760 ms  18.757 ms
6  TenGigabitEthernet9-2.ar1.ATL2.gblx.net (64.212.108.69)  18.751 ms  17.168 ms  17.187 ms
7  The-Planet-Dallas.TenGigabitEthernet6-2.ar5.DAL2.gblx.net (67.17.168.94)  44.432 ms  42.692 ms  42.683 ms
8  te2-4.dsr01.dllstx3.networklayer.com (70.87.255.38)  42.643 ms  50.453 ms  50.373 ms
9  * * *
10  a.ff.5746.static.theplanet.com (70.87.255.10)  46.735 ms  57.331 ms  57.198 ms
11  8c.bb.78ae.static.theplanet.com (174.120.187.140)  57.064 ms  43.144 ms  43.161 ms
[user@clearworld1 ~]# tracert 174.120.187.140
traceroute to 174.120.187.140 (174.120.187.140), 30 hops max, 40 byte packets
1  10.241.192.1 (10.241.192.1)  7.119 ms  10.391 ms  10.440 ms
2  gi15-0-0-3101.gnboncsg-rtr2.triad.rr.com (24.28.225.189)  11.872 ms  11.923 ms  11.918 ms
3  24.93.64.76 (24.93.64.76)  18.758 ms  18.974 ms  19.255 ms
4  107.14.19.18 (107.14.19.18)  19.327 ms  20.083 ms  20.136 ms
5  107.14.19.11 (107.14.19.11)  20.122 ms  20.155 ms  20.157 ms
6  TenGigabitEthernet9-2.ar1.ATL2.gblx.net (64.212.108.69)  20.147 ms  19.157 ms  19.913 ms
7  The-Planet-Dallas.TenGigabitEthernet6-2.ar5.DAL2.gblx.net (67.17.168.94)  47.068 ms  49.080 ms  49.076 ms
8  te2-4.dsr01.dllstx3.networklayer.com (70.87.255.38)  203.466 ms  186.419 ms  186.297 ms
9  * * *
10  a.ff.5746.static.theplanet.com (70.87.255.10)  46.456 ms  46.429 ms  44.629 ms
11  8c.bb.78ae.static.theplanet.com (174.120.187.140)  44.455 ms  46.665 ms  46.658 ms
[user@clearworld1 ~]# tracert 174.120.187.140
traceroute to 174.120.187.140 (174.120.187.140), 30 hops max, 40 byte packets
1  10.241.192.1 (10.241.192.1)  10.905 ms  11.650 ms  11.697 ms
2  gi15-0-0-3101.gnboncsg-rtr2.triad.rr.com (24.28.225.189)  11.693 ms  11.685 ms  11.670 ms
3  24.93.64.76 (24.93.64.76)  29.820 ms  29.860 ms  29.856 ms
4  107.14.19.18 (107.14.19.18)  29.839 ms  29.824 ms  29.800 ms
5  107.14.19.11 (107.14.19.11)  29.772 ms  29.751 ms  29.727 ms
6  TenGigabitEthernet9-2.ar1.ATL2.gblx.net (64.212.108.69)  29.847 ms  172.769 ms  172.683 ms
7  The-Planet-Dallas.TenGigabitEthernet6-2.ar5.DAL2.gblx.net (67.17.168.94)  49.454 ms  53.293 ms  53.279 ms
8  te2-4.dsr01.dllstx3.networklayer.com (70.87.255.38)  53.254 ms  241.074 ms  241.300 ms
9  * * *
10  a.ff.5746.static.theplanet.com (70.87.255.10)  53.405 ms  54.124 ms  54.152 ms
11  8c.bb.78ae.static.theplanet.com (174.120.187.140)  54.065 ms  52.321 ms  52.329 ms
[user@clearworld1 ~]# tracert 174.120.187.140
traceroute to 174.120.187.140 (174.120.187.140), 30 hops max, 40 byte packets
1  10.241.192.1 (10.241.192.1)  6.987 ms  6.965 ms  6.999 ms
2  gi15-0-0-3101.gnboncsg-rtr2.triad.rr.com (24.28.225.189)  11.634 ms  11.692 ms  12.094 ms
3  24.93.64.76 (24.93.64.76)  14.538 ms  14.685 ms  14.735 ms
4  107.14.19.18 (107.14.19.18)  24.952 ms  24.960 ms  24.953 ms
5  107.14.19.11 (107.14.19.11)  24.943 ms  24.927 ms  24.915 ms
6  TenGigabitEthernet9-2.ar1.ATL2.gblx.net (64.212.108.69)  27.056 ms  18.807 ms  18.721 ms
7  The-Planet-Dallas.TenGigabitEthernet6-2.ar5.DAL2.gblx.net (67.17.168.94)  45.510 ms  53.255 ms  53.631 ms
8  te2-4.dsr01.dllstx3.networklayer.com (70.87.255.38)  236.022 ms  220.118 ms  220.042 ms
9  * * *
10  a.ff.5746.static.theplanet.com (70.87.255.10)  50.378 ms  45.848 ms  45.679 ms
11  8c.bb.78ae.static.theplanet.com (174.120.187.140)  45.525 ms  55.311 ms  55.301 ms
[user@clearworld1 ~]# tracert 174.120.187.140
traceroute to 174.120.187.140 (174.120.187.140), 30 hops max, 40 byte packets
1  10.241.192.1 (10.241.192.1)  16.556 ms  16.450 ms  16.406 ms
2  gi15-0-0-3101.gnboncsg-rtr2.triad.rr.com (24.28.225.189)  16.362 ms  16.336 ms  16.330 ms
3  24.93.64.76 (24.93.64.76)  16.301 ms  16.280 ms  16.271 ms
4  107.14.19.18 (107.14.19.18)  26.623 ms  26.680 ms  26.674 ms
5  107.14.19.11 (107.14.19.11)  17.784 ms  17.846 ms  17.841 ms
6  TenGigabitEthernet9-2.ar1.ATL2.gblx.net (64.212.108.69)  20.678 ms  17.903 ms  17.901 ms
7  The-Planet-Dallas.TenGigabitEthernet6-2.ar5.DAL2.gblx.net (67.17.168.94)  45.089 ms  45.146 ms  45.121 ms
8  te2-4.dsr01.dllstx3.networklayer.com (70.87.255.38)  45.099 ms  45.090 ms  45.040 ms
9  * * *
10  a.ff.5746.static.theplanet.com (70.87.255.10)  48.017 ms  48.181 ms  45.803 ms
11  8c.bb.78ae.static.theplanet.com (174.120.187.140)  42.230 ms  44.932 ms  44.914 ms
[user@clearworld1 ~]# tracert 174.120.187.140
traceroute to 174.120.187.140 (174.120.187.140), 30 hops max, 40 byte packets
1  10.241.192.1 (10.241.192.1)  14.848 ms  14.824 ms  14.825 ms
2  gi15-0-0-3101.gnboncsg-rtr2.triad.rr.com (24.28.225.189)  16.225 ms  16.258 ms  16.257 ms
3  24.93.64.76 (24.93.64.76)  18.648 ms  18.693 ms  18.687 ms
4  107.14.19.18 (107.14.19.18)  24.013 ms  24.065 ms  24.062 ms
5  107.14.19.11 (107.14.19.11)  24.052 ms  24.048 ms  24.045 ms
6  TenGigabitEthernet9-2.ar1.ATL2.gblx.net (64.212.108.69)  24.234 ms  16.609 ms  16.590 ms
7  The-Planet-Dallas.TenGigabitEthernet6-2.ar5.DAL2.gblx.net (67.17.168.94)  42.949 ms  48.901 ms  48.942 ms
8  te2-4.dsr01.dllstx3.networklayer.com (70.87.255.38)  48.851 ms  47.632 ms  47.626 ms
9  * * *
10  a.ff.5746.static.theplanet.com (70.87.255.10)  48.592 ms  56.616 ms  56.535 ms
11  8c.bb.78ae.static.theplanet.com (174.120.187.140)  56.434 ms  42.600 ms  45.919 ms
[user@clearworld1 ~]#

[TriRan]

atm i think i'm having a peering issue aswell, i have cox 50Mbit line the same type of line CA3LE has and i'm only getting about 25Mbit results from testmy.net but i just downloaded a full server backup from my server in chicago at 7.8MB/s (62.4Mbps) my upstream is unaffected

my tracert looks just fine i posted one earlier in another post so its definately not a routing issue for me

[mudmanc4]

Iv'e been working on the same issue for a while now myself. Much of my traffic is routed through D.C. before heading to it's intended destination. Not as rule. None the less , I'm on a 50/3 connection and have issues much like yourself. It's been this way for roughly 18 months give or take.

Don't quote me on this , because it's merely speculation, I believe the countries networks have been divided in some way on a large scale , possibly for security reasons. And there simply not finished re routing traffic properly due to this filtering. Or cannot find a sane way to detect the type of traffic there looking for. Again just speculation.

At least what I gather is four major sectors , quite possibly five but you have to take notice at Limelight networks as well in california ( hosting Groupon , Netflix , I believe MLB network just to name a few biggies) , divided at the Mississippi , somewhere along the mason dixon line longitudinally , leaving california and upper USA , somewhat resembling the map below. Again I'm reaching here , but something is def up !

[CA3LE]

TriRan, on 03 October 2011 - 12:01 AM, said:

atm i think i'm having a peering issue aswell, i have cox 50Mbit line the same type of line CA3LE has and i'm only getting about 25Mbit results from testmy.net but i just downloaded a full server backup from my server in chicago at 7.8MB/s (62.4Mbps) my upstream is unaffected

my tracert looks just fine i posted one earlier in another post so its definately not a routing issue for me

You know Cox in our area is quoting Ultimate as 55Mbps / 5.5Mbps not 50/5 now. --- at least that's what came up on Cox's website the other day.

I think I'll be throwing a second server online so that people can have a second TestMy.net location to test to... to help clear up any question to routing issues.  Maybe I'll put it to a vote what city I put it in.  I definitely want to put it far away from the main server. ... hummm

Hop 8 doesn't seem to be responding correctly in a couple of those tracerts -- that may be within my host.

... 8 te2-4.dsr01.dllstx3.networklayer.com (70.87.255.38) 236.022 ms 220.118 ms 220.042 ms

I'll let the datacenter know

[TriRan]

CA3LE, on 03 October 2011 - 10:50 AM, said:

You know Cox in our area is quoting Ultimate as 55Mbps / 5.5Mbps not 50/5 now. --- at least that's what came up on Cox's website the other day.

I think I'll be throwing a second server online so that people can have a second TestMy.net location to test to... to help clear up any question to routing issues.  Maybe I'll put it to a vote what city I put it in.  I definitely want to put it far away from the main server. ... hummm

Hop 8 doesn't seem to be responding correctly in a couple of those tracerts -- that may be within my host.

... 8 te2-4.dsr01.dllstx3.networklayer.com (70.87.255.38) 236.022 ms 220.118 ms 220.042 ms

I'll let the datacenter know

i think chicago would be another good location, they have very well developed datacenters there thats the 2nd best location in the US that i can think of

[CA3LE]

[update]
Opened a ticket with the datacenter to make sure there isn't any routing issues that I have control over.  ... I'll let you know when they update me.

[CA3LE]

TriRan, on 03 October 2011 - 10:54 AM, said:

i think chicago would be another good location, they have very well developed datacenters there thats the 2nd best location in the US that i can think of

I was thinking the same thing.  It's a big expense but I think it's about time to add another server.  I think having a second location to another major internet hub will clear up ANY confusion about route.  



I only use dedicated servers, so I'm sorry that I haven't put more online yet.  But if you want the quality of test that I strive for you HAVE to go dedicated.  VPS doesn't cut it, too many variables from other people on the server... makes for a very bad platform.  The way that I do it is more expensive but it ensures that no BS is bleeding into the results.  ---- I can guarantee you that the other guys don't care as much about stuff like that.  I could have put 100 shitty VPS servers along time ago.  Trust me, I explored the idea and it's well within the sites budget but I feel it's not the right way.  The only way to be sure about the variables is to have full control over them.  Dedicated... and then if you go dedicated you might as well get at least 8 cores, dual gigabit... etc.  

I kinda want to just order it up... I've been thinking about it for a long time.  hummm

[mudmanc4]

Have you looked into Washington D.C. ? Or even Cleveland

[CA3LE]

mudmanc4, on 03 October 2011 - 11:54 AM, said:

Have you looked into Washington D.C. ? Or even Cleveland

D.C. isn't bad

CA3LE, on 03 October 2011 - 10:55 AM, said:

[update]
Opened a ticket with the datacenter to make sure there isn't any routing issues that I have control over.  ... I'll let you know when they update me.

Still waiting on a response...

[CA3LE]

Here's the bandwidth to the various datacenters I work with

http://www.softlayer...etwork/carriers

Washington D.C. has some sick bandwidth...

[mudmanc4]

4x10 Gbps , nothing to scoff at. Order it up and have them run dedicated fiber to the north side of my house

YEA , I know it's gigabit , amazing

[CA3LE]

... I'm just waiting for them to get the new server online.  I got it in D.C. -- they have PoP through Chicago but no datacenter.  They just opened datacenters in Amsterdam and Singapore so after I get this configured we'll see about opening up some stuff for the folks over seas.  

... it will take me some time to get this configured.

[mudmanc4]

I've been waiting for this for a long time dude !


Not only that , I believe you just made a vital move towards the expansion and social trust of testmy.net ,imo , you could not have made a better move , and D.C. to boot !

[TriRan]

CA3LE, on 03 October 2011 - 01:11 PM, said:

... I'm just waiting for them to get the new server online.  I got it in D.C. -- they have PoP through Chicago but no datacenter.  They just opened datacenters in Amsterdam and Singapore so after I get this configured we'll see about opening up some stuff for the folks over seas.  

... it will take me some time to get this configured.

yeah dc would've been my 3rd choice they aren't quite as centrally located but have amazing peering

[CA3LE]

This new server I'm setting up is sweet



http://testmy.net/st...=25&z=8&q=CA3LE

Just messing around in VNC before any real configuration.

mudmanc4, on 03 October 2011 - 01:26 PM, said:

I've been waiting for this for a long time dude !


Not only that , I believe you just made a vital move towards the expansion and social trust of testmy.net ,imo , you could not have made a better move , and D.C. to boot !

I'm already planning Amsterdam and Singapore after I'm done configuring this one.  I worked a deal with my host.  So folks in Europe and Asia will be happy to hear that.  

[mudmanc4]

Tight as an eagles ass in a power dive !!!!!!!!!!

[CA3LE]

I'm not done configuring but if you guys want to give it a try ... http://dc.testmy.net/

[TriRan]

right on man,


DC


Dallas

[CA3LE]

TriRan, on 03 October 2011 - 10:38 PM, said:

right on man,


DC


Dallas

So weird that you'd have a routing issue to Dallas... when I'm in the same city... same ISP and I'm not seeing an issue.  The internet is weird.  At least we now have a second server to test with...

Man, see what you started thomasrelee... you make a post and less than 24 hours after I read it TMN's got an extra server online.  

D.C. has some sweet bandwidth too, very happy with the new server

[thomasrelee]

CA3LE, on 04 October 2011 - 01:00 AM, said:

So weird that you'd have a routing issue to Dallas... when I'm in the same city... same ISP and I'm not seeing an issue.  The internet is weird.  At least we now have a second server to test with...

Man, see what you started thomasrelee... you make a post and less than 24 hours after I read it TMN's got an extra server online.  

D.C. has some sweet bandwidth too, very happy with the new server

Not bad at all - but you did not have to go through all of this... Thank you!!!  That is the least I can say!



Almost always testing against the DC server, the tests are over my lines rated speed of 15 mbps, averaging 17.35 mbps for 11 randomly run tests over the past three hours!!!

One or two of the tests somehow were launched from the Dallas server ( I guess it is due to some configurations you are still working on, and I did not include them in the above average ), but DC seems like they have their routing/peering in great shape.  As for the tests that hit the Dallas server, they were still well below my lines rated speed.

I hope your host finds whatever the bug is for your Dallas server and squashes it.