Charting Fibre speed over time.

[jacof]

As you can see no matter the ISP I use I follow the same path up to the BNG and from that point I am on the ISP network.

"Dit is wanneer n search gehardloop word op jou physical port wat altyd konstant bly. IP van ESR/BNG bly dieselfde. Jou circuit nommer bly dieselfde, VLANs bly dieselfde en al wat verskil is die ISP accounts, maar almal hardloop dieselfde pad to by die BNG. Die BNG direct dan die traffic (BNG = jou 2de Hop) na die ISP en dan is dit in ISP domain by die 3de hop en buite OS net soos jou tracert dit bewys."

Thus how can this be on the OS side. I must admit OS has been very helpful to show it is not on their end. The ISP in general don't even answer. Paul from CI have been helpful as well but I have not seen proof from Cool Ideas that there is no issue on their end other than a IPC chart.

So for me it is clear at this stage. I will be going forward using VOX over Openserve, that is the most stable option for me at this stage.

 

[MickeyD]

We have no reason not to believe the impacted ISPs when they rule out capacity issues on their side. Other than them and OS, who else has a part to play in the total supply chain?

 

[PBCool]

I don't think you understand the way layer2 works, it's not IP related which is what you are posting. The layer2 is "underneath" which is the path from your house to our handover. Nothing to do with a trace.

 

[PBCool]

We have no reason not to believe the impacted ISPs when they rule out capacity issues on their side. Other than them and OS, who else has a part to play in the total supply chain?

No one, I have supplied my IPC graph to Jaco and his graph dips when ours actually is under utilised. We are engaging with OS to try an troubleshoot what is happening at layer2, we have no doubt there is a problem so we appreciate the feedback and will get it resolved for you.

 

[PBCool]

As you can see no matter the ISP I use I follow the same path up to the BNG and from that point I am on the ISP network.

"Dit is wanneer n search gehardloop word op jou physical port wat altyd konstant bly. IP van ESR/BNG bly dieselfde. Jou circuit nommer bly dieselfde, VLANs bly dieselfde en al wat verskil is die ISP accounts, maar almal hardloop dieselfde pad to by die BNG. Die BNG direct dan die traffic (BNG = jou 2de Hop) na die ISP en dan is dit in ISP domain by die 3de hop en buite OS net soos jou tracert dit bewys."

Thus how can this be on the OS side. I must admit OS has been very helpful to show it is not on their end. The ISP in general don't even answer. Paul from CI have been helpful as well but I have not seen proof from Cool Ideas that there is no issue on their end other than a IPC chart.

So for me it is clear at this stage. I will be going forward using VOX over Openserve, that is the most stable option for me at this stage.

If it's not IPC then have someone run the same graphs you are on another network? I have no doubt it's OS and IPC related.

 

[jacof]

If it's not IPC then have someone run the same graphs you are on another network? I have no doubt it's OS and IPC related.

Who is willing to do the tests? I will help setup the code etc. I just need a pc on another carrier other than Openserve. We have Test accounts to use.

 

[PBCool]

Who is willing to do the tests? I will help setup the code etc. I just need a pc on another carrier other than Openserve. We have Test accounts to use.

Just a heads up it's not so easy on the other networks so best to get a few people on the respective ISPs . You can't just switch accounts.

 

[gfmalan]

Who is willing to do the tests? I will help setup the code etc. I just need a pc on another carrier other than Openserve. We have Test accounts to use.

I'm on Octotel and will make whatever available!

 

[PBCool]

I'm on Octotel and will make whatever available!

Is great to see the community piece together to do these kind of experiments .

 

[Armizael]

I was going to offer to run these tests as well (since I'm on Frogfoot), but seems like there's something wrong when I test to the CI server.

Using the CI Randburg server:

http://beta.speedtest.net/result/6548055579

All the others seem fine-ish though:

http://beta.speedtest.net/result/6548100323.png
http://beta.speedtest.net/result/6548105770.png
http://beta.speedtest.net/result/6548109674.png

@PBCool I'll open a ticket for this - if we can get my speedtests sorted to CI servers, then I don't mind participating.
http://beta.speedtest.net/result/6548097644.png

 

[PBCool]

I was going to offer to run these tests as well (since I'm on Frogfoot), but seems like there's something wrong when I test to the CI server.

Using the CI Randburg server:

http://beta.speedtest.net/result/6548055579

All the others seem fine-ish though:

http://beta.speedtest.net/result/6548100323.png
http://beta.speedtest.net/result/6548105770.png
http://beta.speedtest.net/result/6548109674.png

@PBCool I'll open a ticket for this - if we can get my speedtests sorted to CI servers, then I don't mind participating.
http://beta.speedtest.net/result/6548097644.png

This over cable? Latency is all over the place.

 

[Armizael]

Yip - over cable.
It's weird that the latency is all over the place with the speed tests - pinging bbc.co.uk yields a fairly stable result:

Ping statistics for 212.58.246.93:
Packets: Sent = 74, Received = 74, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 165ms, Maximum = 168ms, Average = 166ms

I opened a ticket though if you want to troubleshoot - COOL-20170817-31171

Edit: Looks like it's fixed now - weird.
http://beta.speedtest.net/result/6548185300.png

I'll set up a script to run a speedtest to the Randburg CI server every 5 minutes and save it in a CSV - can always graph it out afterwards.

 

[jacof]

We have now set-up a PC at gfmalan and running the same test but not over the Openserve fibre but over the Octotel network. We are seeing the same drops in speed to the Cool Ideas ookla server in Randburg. We will let it run for the day and then we will switch ISP to do the test again. This is a 100 / 100 Mbps line but the upload is below 20 Mbps.

Here are the live charts.

https://emoncms.org/jacof/testsite2

 

[PBCool]

We have now set-up a PC at gfmalan and running the same test but not over the Openserve fibre but over the Octotel network. We are seeing the same drops in speed to the Cool Ideas ookla server in Randburg. We will let it run for the day and then we will switch ISP to do the test again. This is a 100 / 100 Mbps line but the upload is below 20 Mbps.

Here are the live charts.

https://emoncms.org/jacof/testsite2

Gfmalan you on our network? There is no way those graphs are normal. Also remember as I said you can just switch ISP on the Octotel network. It would have to be another person to test via another provider.

 

[PBCool]

Yip - over cable.
It's weird that the latency is all over the place with the speed tests - pinging bbc.co.uk yields a fairly stable result:

Ping statistics for 212.58.246.93:
Packets: Sent = 74, Received = 74, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 165ms, Maximum = 168ms, Average = 166ms

I opened a ticket though if you want to troubleshoot - COOL-20170817-31171

Edit: Looks like it's fixed now - weird.
http://beta.speedtest.net/result/6548185300.png

I'll set up a script to run a speedtest to the Randburg CI server every 5 minutes and save it in a CSV - can always graph it out afterwards.

Yeah more than likely still Frogfoot having issues.

 

[gfmalan]

Gfmalan you on our network? There is no way that ping graph is correct. Also remember as I said you can just switch ISP on the Octotel network. It would have to be another person to test via another provider.

I am, I gave your T1 helpdesk guy access todo his own tests and he found similar issues.

 

[PBCool]

I am, I gave your T1 helpdesk guy access todo his own tests and he found similar issues.

Cool will have a look, will you PM me your account number? Have the tech logged it with Octotel?

 

[gfmalan]

Cool will have a look, will you PM me your account number? Have the tech logged it with Octotel?

According to the ticket and conversation yes, and it took from a couple of days and Octotel said no problem found.

I can setup whatever you need todo tests, I also have ADSL line so you can connect to my network and leave the fiber side just for your tests.

I'll PM you

 

[PBCool]

According to the ticket and conversation yes, and it took from a couple of days and Octotel said no problem found.

I can setup whatever you need todo tests, I also have ADSL line so you can connect to my network and leave the fiber side just for your tests.

I'll PM you

Cool Octotel has had a general fault the past couple of days so they may be talking about your line specifically not their metro which is where they are having issues.

 

[jacof]

@gfmalan it is your call but can we let the test run for the day. We now have 3 test sites showing similar issues using diffrent carriers. The data is starting to form a picture.