Cool Ideas Fibre ISP - Feedback

[John Tempus]

what is the ip address of your vpn? If you dont mind, showing us a traceroute to the vpn? I suspect the vpn is on one of the free peers at LINX. I also suspect the slow torrents could be because of something else like HE.net but who knows.

VPN is hosted in USA New York. This is not a local south african VPN so connecting to it still requires stable bandwidth from SA to USA New York. Id prefer not to publish the VPN ip as it is part of a public subnet of my company and better to keep it out of public eye.

I can show you mtr results up until the hop before my host. What the hell is up with that he.ixp.joburg ping of 164ms but that aside going over this route to my VPN I am getting reliable download speed on all protocols. Just http response is a bit slower compared to no VPN due to the higher ping between every request.

Funny you mention slow torrents due to HE.net and lobehold my VPN route is going right over he.net and not having any issues over the VPN, haha.

|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| 192.168.0.1 - 0 | 19 | 19 | 0 | 0 | 1 | 0 |
| 155.93.248.1 - 0 | 19 | 19 | 1 | 1 | 15 | 1 |
| c3h-backbone.coolideas.co.za - 0 | 19 | 19 | 1 | 1 | 2 | 1 |
| cd-backbone.coolideas.co.za - 0 | 19 | 19 | 1 | 1 | 2 | 1 |
| c6t-backbone.coolideas.co.za - 0 | 19 | 19 | 20 | 20 | 22 | 21 |
| c3x-backbone.coolideas.co.za - 0 | 19 | 19 | 21 | 21 | 21 | 21 |
| he.ixp.joburg - 0 | 19 | 19 | 162 | 164 | 173 | 172 |
| 10ge3-11.core1.lon2.he.net - 0 | 19 | 19 | 162 | 168 | 216 | 163 |
| 100ge4-1.core1.nyc4.he.net - 0 | 19 | 19 | 230 | 234 | 241 | 237 |
|firstlight-fiber.10gigabitethernet7-3.core1.nyc4.he.net - 0 | 19 | 19 | 230 | 230 | 233 | 231 |
| 66-152-98-9.tvc-ip.com - 0 | 19 | 19 | 235 | 235 | 237 | 236 |
| 66-109-52-166.tvc-ip.com - 0 | 19 | 19 | 233 | 235 | 285 | 233 |
| xe-2-3-0.edge2.ny01.as40244.net - 0 | 19 | 19 | 235 | 237 | 258 | 235 |
| Packet was too big. - 100 | 19 | 0 | 0 | 0 | 0 | 0 |
| xe-1-0-1.core107.ny01.as40244.net - 0 | 19 | 19 | 235 | 235 | 236 | 235 |

 

[Tinuva]

VPN is hosted in USA New York. This is not a local south african VPN so connecting to it still requires stable bandwidth from SA to USA New York. Id prefer not to publish the VPN ip as it is part of a public subnet of my company and better to keep it out of public eye.

Exactly, its going through international, so there are still lots of variables to compare. Feel free to send that details to CISP at least. Either way, posting it here doesnt make it unsafe, your company's security on the subnet will make it unsafe if its lacking

 

[John Tempus]

Exactly, its going through international, so there are still lots of variables to compare. Feel free to send that details to CISP at least. Either way, posting it here doesnt make it unsafe, your company's security on the subnet will make it unsafe if its lacking

I shouldnt need to provide IP's in any case of remote services outside coolideas. The info I provided here and the comparisons speak for themself.

Just to sweeten the situation I even setup a torrent on my VPN host to test speed to it without connected to the VPN, same BS slow speed and the moment I connect to the VPN and download the torrent from the same host the speed issues goes away. This must be the biggest case of coincidences I have ever witnessed or I can go by facts and point out that when I connect and download from my own server without connecting to the VPN thats hosted on the exact same same the speed is next to useless. Switch on the VPN connection hosted on the same server and the torrent test from the same server is running along perfectly fine.

You should know that avoiding to publish your public ips if possible is a far better security approach than to willy nilly post it in a public forum but thats besides the point of everything that I posted here regarding international performance issues.

I cant be the only one able to replicate these same scenarios and speed results so I will silently wait here while others start to hopefully report their own results and in the meantime just keep using my VPN since that is the only solution for reliable international speed.

This is my last post about this because it doesn't seem to me any progress is being made or real transparency about coolideas setup is given to us and I am simply wasting my energy.

 

[Bedford]

I did some testing now on p2p.

Downloading an Ubuntu distro over torrent, i was able to max out my line in about 1:30min (while keeping a 20kb/s upload limit). Using the same configuration i tried downloading a movie, seemed to max out around 1.4mb/s, but noticed the seed/leech ratio was a little to close together. So I found another movie with a 2:1 ratio of Seed/Leech. After about 7mins of downloading the max i could get was around 1.6mb/s. I took the upload limit off and my download climbed to around 2.3mb/s. I then increased then increased the maximum concurrent peers per torrent to 200 and eventually got to around 3.2mb/s.

So with Ubuntu torrent it maxed out my speed relatively quickly.
With a movie, i had to tweak the uTorrent settings to get more throughout.

Hope that helps.

Speedtest to London:
http://www.speedtest.net/result/7057371594.png
One thing i have noticed, doesn't matter what international ISP i test. My upload is always capped at 5mbps and it never use to be like that.

With speedtest, is it just downloading/uploading a test file. Could it be considered a accurate indication of oversees traffic?

 

[PBCool]

So a couple of things here, we use 2 upstreams in CPT for bandwidth eNetworks/Liquid Telecom for IPT and then we have WACS to London where we pick up our London POP, the same as we do in JHB. Obviously on the layer2 and circuit to London we have full control of but on the IPT side we don't, even though we purchase this as a dedicated service which doesn't entail any shaping.

This isn't a conspiracy where we are trying to shape protocols to save money and oversell bandwidth.

If you tunnel to Germany does your PC get a public IP address?

 

[PBCool]

I shouldnt need to provide IP's in any case of remote services outside coolideas. The info I provided here and the comparisons speak for themself.

Just to sweeten the situation I even setup a torrent on my VPN host to test speed to it without connected to the VPN, same BS slow speed and the moment I connect to the VPN and download the torrent from the same host the speed issues goes away. This must be the biggest case of coincidences I have ever witnessed or I can go by facts and point out that when I connect and download from my own server without connecting to the VPN thats hosted on the exact same same the speed is next to useless. Switch on the VPN connection hosted on the same server and the torrent test from the same server is running along perfectly fine.

You should know that avoiding to publish your public ips if possible is a far better security approach than to willy nilly post it in a public forum but thats besides the point of everything that I posted here regarding international performance issues.

I cant be the only one able to replicate these same scenarios and speed results so I will silently wait here while others start to hopefully report their own results and in the meantime just keep using my VPN since that is the only solution for reliable international speed.

This is my last post about this because it doesn't seem to me any progress is being made or real transparency about coolideas setup is given to us and I am simply wasting my energy.

I'm not sure what you mean by no transparency, I have been completely transparent. So these tests you are now doing during the day?

 

[Tinuva]

I shouldnt need to provide IP's in any case of remote services outside coolideas. The info I provided here and the comparisons speak for themself.

Just to sweeten the situation I even setup a torrent on my VPN host to test speed to it without connected to the VPN, same BS slow speed and the moment I connect to the VPN and download the torrent from the same host the speed issues goes away. This must be the biggest case of coincidences I have ever witnessed or I can go by facts and point out that when I connect and download from my own server without connecting to the VPN thats hosted on the exact same same the speed is next to useless. Switch on the VPN connection hosted on the same server and the torrent test from the same server is running along perfectly fine.

You should know that avoiding to publish your public ips if possible is a far better security approach than to willy nilly post it in a public forum but thats besides the point of everything that I posted here regarding international performance issues.

I cant be the only one able to replicate these same scenarios and speed results so I will silently wait here while others start to hopefully report their own results and in the meantime just keep using my VPN since that is the only solution for reliable international speed.

This is my last post about this because it doesn't seem to me any progress is being made or real transparency about coolideas setup is given to us and I am simply wasting my energy.

First, if you have a public subnet, there are already 1000s of bots that know about it, and scanned it for holes to attack. Not posting it here, makes it impossible for anyone trying to help to know which path on the internet it has followed.

Not even looking at replicating the issue, because well it will be impossible to know which parts of the network to test. It is not as simple as just international. The troubleshooter needs to know to which peer of transit provider the traffic go to and originate from is what will help establish facts from fiction.

Look, I believe, there is something going on here. Its just impossible to find with the current available information.
Knowing torrents looked shaped vs vpn not shaped, over the same path is good information. But without knowing which network the traffic pass over means it will be near impossible to troubleshoot. I would put it on my list of things to check out, but may prioritize a task where I have complete information.

I did mention you should pass on the ip address and info to CISP. You really do want to give them all the information possible to assist you.

 

[John Tempus]

So a couple of things here, we use 2 upstreams in CPT for bandwidth eNetworks/Liquid Telecom for IPT and then we have WACS to London where we pick up our London POP, the same as we do in JHB. Obviously on the layer2 and circuit to London we have full control of but on the IPT side we don't, even though we purchase this as a dedicated service which doesn't entail any shaping.

This isn't a conspiracy where we are trying to shape protocols to save money and oversell bandwidth.

If you tunnel to Germany does your PC get a public IP address?

VPN in germany I used initially were unreliable since its a public service and it would not be a fair assessment. Thats why I configured one on my USA host where I know exactly what bandwidth I have available in order to saturate the connection if indeed possible. This is exactly what happened, over the VPN to my USA, New York host I am able to saturate the connection on anything protocol I test on the same USA , New York host. The moment I disconnect the VPN and I test any protocol traffic to the USA, New York host the same under 1Mbit/s speed returns.

 

[John Tempus]

First, if you have a public subnet, there are already 1000s of bots that know about it, and scanned it for holes to attack. Not posting it here, makes it impossible for anyone trying to help to know which path on the internet it has followed.

Not even looking at replicating the issue, because well it will be impossible to know which parts of the network to test. It is not as simple as just international. The troubleshooter needs to know to which peer of transit provider the traffic go to and originate from is what will help establish facts from fiction.

Look, I believe, there is something going on here. Its just impossible to find with the current available information.
Knowing torrents looked shaped vs vpn not shaped, over the same path is good information. But without knowing which network the traffic pass over means it will be near impossible to troubleshoot. I would put it on my list of things to check out, but may prioritize a task where I have complete information.

I did mention you should pass on the ip address and info to CISP. You really do want to give them all the information possible to assist you.

You seem to be stuck on the wrong issue here. My point is that over my VPN I have no international issues and without the VPN international is as unreliable as a futuristic dialup connection.

Not sure why you are so up in arms about where my host is, I am not trying to resolve performance without the VPN to this USA host. This host I am just using as a benchmark as to why my international is so pisspoor unreliable anywhere without the VPN and the moment I connect to my VPN the international issues goes away.

Again the IP of the host I run my VPN server on have absolutely nothing to do with any of the issues nor will that magically give information about why all the other international traffic is shamefully slow without the VPN connection.

Anyone reading this thread can come to the exact same conclusions. Find a reliable UDP based VPN that you can connect to that you know for certain is not overutilised and not blocking p2p traffic. Now find a reliable torrent that you know have thousands of seeds and replicate this test. You will also find that the speed with the VPN far outpace that of not using the VPN. The nordvpn i initially used that I could not reliably say what bandwidth it had available also showed the same huge difference so I find it impossible that I am only the one able to replicate this over and over.

 

[Tinuva]

You seem to be stuck on the wrong issue here. My point is that over my VPN I have no international issues and without the VPN international is as unreliable as a futuristic dialup connection.

Not sure why you are so up in arms about where my host is, I am not trying to resolve performance without the VPN to this USA host. This host I am just using as a benchmark as to why my international is so pisspoor unreliable anywhere without the VPN and the moment I connect to my VPN the international issues goes away.

Again the IP of the host I run my VPN server on have absolutely nothing to do with any of the issues nor will that magically give information about why all the other international traffic is shamefully slow without the VPN connection.

Anyone reading this thread can come to the exact same conclusions. Find a reliable UDP based VPN that you can connect to that you know for certain is not overutilised and not blocking p2p traffic. Now find a reliable torrent that you know have thousands of seeds and replicate this test. You will also find that the speed with the VPN far outpace that of not using the VPN. The nordvpn i initially used that I could not reliably say what bandwidth it had available also showed the same huge difference so I find it impossible that I am only the one able to replicate this over and over.

You missed the part where I did say that is good info, but on its own its not easy to pin point where this shaping is happening.
I get your point, but that is all.

Do you want to figure out where this shaping happen? I'd say from here your options are, to do the same test to other international locations on different routes. Still, it will be difficult to tell where the issue is. I suspect its a specific international provider, but its only a guess until there are traceroutes, which either you do or others do by knowing what the ips are. Its more about the path to the ip than it is about the ip, which you are stuck on supplying.

Imagine this. You are on a road, with still standing cars not moving. You know up ahead the road split into 4 roads. 1 or more of those roads are slow, the others are fast. Do you know which ones are slow without knowing the destination?

 

[eddief1]

@John Tempus....Just clutching at straws here.

1) Have you tried a TCP based VPN? If so are you still running at full speed? IE the problem could be TCP window scaling which does not happen when you tunneling all your traffic through UDP

 

[Y'all Qaeda]

To be fair though, I don't imagine there is anything sinister going on and purely by judging the response from PBCool, I can't flippen wait to stop my current provider.

 

[John Tempus]

@John Tempus....Just clutching at straws here.

1) Have you tried a TCP based VPN? If so are you still running at full speed? IE the problem could be TCP window scaling which does not happen when you tunneling all your traffic through UDP

I know its easy to miss in all the text in my previous comments. I tested random tcp ports, common and uncommon and they were all fked and speed all over the place. Slightly better than no VPN but the difference were small enough not to matter.

TCP window scaling will be a non issue when you connect to 2000+ private and reliable seed peers. Even if it were just a single reliably fast seed peer TCP scaling would only matter for the first minute or so.

UDP were from the get go much better but not all UDP ports. I eventually narrowed down an obscure UDP port that seem to be the most reliable and that is the one I am now using for all my international traffic.

 

[John Tempus]

To be fair though, I don't imagine there is anything sinister going on and purely by judging the response from PBCool, I can't flippen wait to stop my current provider.

I know and trust and want to believe it is something seriously obscure causing this. But at the same time I have gone down this rabbithole with so many other providers in the past where I felt I was receiving all honest information just to get the inside scoop later on that similar situational information were not honestly given to the public.

That said I still want to trust that PBCool is on the up and up on everything that he personally knows, I dont want him to either speculate or guess about things he might not have information about.

But I need to face my evidence and the responses on my past experiences dealing with ISP's and even the most honest and well known people I dealt with either didnt know all the information about their networks/backbone providers or chose to not provide the full truth when it suited them.

Just to add some extra information. On the VPN everything except http traffic can easily max out my connection. Disconnecting the VPN my http traffic seems to nearly max out my connection and all other protocols are down the dumps. This difference can only make sense if there is some form of caching involved with http traffic and coolideas without the VPN which then gets bypassed when I connect to the VPN explaining the slower but more consistent speed over the VPN For http traffic ? My mind is just blown at this point.

 

[PBCool]

I know and trust and want to believe it is something seriously obscure causing this. But at the same time I have gone down this rabbithole with so many other providers in the past where I felt I was receiving all honest information just to get the inside scoop later on that similar situational information were not honestly given to the public.

That said I still want to trust that PBCool is on the up and up on everything that he personally knows, I dont want him to either speculate or guess about things he might not have information about.

But I need to face my evidence and the responses on my past experiences dealing with ISP's and even the most honest and well known people I dealt with either didnt know all the information about their networks/backbone providers or chose to not provide the full truth when it suited them.

Just to add some extra information. On the VPN everything except http traffic can easily max out my connection. Disconnecting the VPN my http traffic seems to nearly max out my connection and all other protocols are down the dumps. This difference can only make sense if there is some form of caching involved with http traffic and coolideas without the VPN which then gets bypassed when I connect to the VPN explaining the slower but more consistent speed over the VPN For http traffic ? My mind is just blown at this point.

The only thing that could be causing this would be our IPT from eNetworks, it would be affecting everyone though.

And just as a heads up it's not about what I know personally, I commission, configure and manage the network so you can believe what I am telling you.

 

[mugikhan]

@PBCool getting lots of packet loss right now...More issues with fibre that I was assured I wouldn't have...

 

[PBCool]

@PBCool getting lots of packet loss right now...More issues with fibre that I was assured I wouldn't have...

Can you give me the example? MTR?

 

[John Tempus]

The only thing that could be causing this would be our IPT from eNetworks, it would be affecting everyone though.

And just as a heads up it's not about what I know personally, I commission, configure and manage the network so you can believe what I am telling you.

I was trying to give you benefit of the doubt if you did not have full control. After what you stated now having full control the issue is even more frustrating that I can clearly see difference over same exact hops with/without VPN showing vastly different results.

There is nothing else I can eliminate, literally comparing apples with apples and getting massively different results.

 

[PBCool]

I was trying to give you benefit of the doubt if you did not have full control. After what you stated now having full control the issue is even more frustrating that I can clearly see difference over same exact hops with/without VPN showing vastly different results.

There is nothing else I can eliminate, literally comparing apples with apples and getting massively different results.

I don't see how that makes sense, we don't have control of "everything" I am being honest about our portion.

Please clarify a couple things for me, at first you stated it was only at peak times an issue.

So during normal hours everything is fine and in the evening if you use the VPN it is also fine?

 

[Tinuva]

I was trying to give you benefit of the doubt if you did not have full control. After what you stated now having full control the issue is even more frustrating that I can clearly see difference over same exact hops with/without VPN showing vastly different results.

There is nothing else I can eliminate, literally comparing apples with apples and getting massively different results.

Are you willing to PM or use another secure method to supply subnet/ip address for further investigation/troubleshooting? (keep in mind I don't work for CISP, I am a huge fan and just want to see actual issues resolves)