Cool Ideas Fibre ISP - Feedback

[wingnut771]

Agreed, I’m on Frogfoot with CISP 100/100 in Sunningdale, Cape Town and very happy. Streaming since July 2018 has been absolutely perfect and haven’t had a single minute downtime besides scheduled maintenance.

As you say, single threaded downloads the only problem, but a small problem compared to some other fibre networks I’ve seen.

My Dropbox performance is also quite good, around 5MB/sec uploads which is about half my line speed.

This is relative. If you were trying to use the internet where single thread performance is the only option, then its a huge problem, I would then cancel and use LTE.

 

[Phinix]

Please fix Twitch. Only on CI does it buffer at peak times.

 

[Wasabee!]

So as I see it we have the following issues that unfortunately CISP is caught in the middle with

1. FTTH providers and flaky (cheap) equipment- Yes Octotel, you and your DASAN ONT and its iffy WAN port
2. Flaky FTTH providers (Frogfoot, and it appears, VUMA)

1. Please explain how the open access network (OAN) is flaky? (I'm picking up a trend of misinformation on the infrastructure provider being to blame be it Vuma, FF, Openserve etc which I'd like to correct.)
2. The OAN provide only layer 2 connectivity from your premises to the regional DC. In Gauteng it's in Teraco Isando where the vast majority of ISPs peer etc. See my previous post for context on how ISPs build their service. In the Western Cape it is Teraco Great Weatherford in Rondebosch. Most of the INX's have become too expensive and Teraco have taken most peering over in terms of volume imo.
3. In reference to point 2 above, the OAN like Vuma only provides layer 2 from your premises to the regional DC. While I'm not clued up with everyone's symptoms etc, I'm confident to assume the issue is with your ISP as your ISP does A. layer 3 connectivity and B. connectivity from the regional DC to the rest of the world. So if you have any international issue, the high probability is it's your ISP (single or multiple threaded problems).
4. The ISP buys IPC from an ISP, often different to the last mile provider/OAN. So what seems to be happening here is that the ISP (Cool Ideas) are congested on their international bandwidth, they just tell their clients oh it's Vuma or Frogfoot. This is passing the buck and is poor ethics. If your local speeds/latency is fine and your international is not, chances are that it is your ISP that need to sort out their capacity or routing etc.

My experience with Frogfoot was the opposite. I told them SADV was useless and never pitched to even do a site survey at the townhouse complex. They literally arranged a meeting with me and the HOA chairperson for the next day, showed up and everything went smooth from thereon.

Only issue now is the slow international speeds with Frogfoot + CISP which is apparently being attended to.

Please see my response above, it sounds like your ISP is the problem not the OAN.

I am on Cisp 100/100 on frogfoot, lynwood pta.

Only issue I have is single threaded download speeds international. Sadly an issue that isn't isolated to me, paul is on it with a few of us in this thread. Sadly FF has been extremely slow to rectify it.

For gaming,streaming,ect I have had no issues. Have hardly had a minutes down time since they installed july last year.

Another example, international is your ISP not the infrastructure provider. If your ISP can't sort it out with the infrastructure provider, get a new ISP that knows what they are doing or doesn't just pass the buck to the OAN.

The fact that some peeps have been having this issue for ~6 months+ on frogfoot does not instill much confidence...

Another example, please see above response.

I can't get 10Mbps on multithread to London with my 20/20 with Frogfoot 90% of the time. Worse to other international servers. Here's a single threaded test to London
View attachment 601544

Another example. Please see above response.

 

[ryanrich]

This is relative. If you were trying to use the internet where single thread performance is the only option, then its a huge problem, I would then cancel and use LTE.

Very true.

 

[image132]

@PBCool

Twitch is not working tonight. Its pulling 180KB/s to 220KB/s.

100/25 Octotel. Cape Town.

London Coreix:

Multi-threaded
https://www.speedtest.net/result/7946658184

Single-threaded
https://www.speedtest.net/result/7946662510

Is there an issue or this my congestion coming back again because holidays are over.... This is driving me nuts.

 

[Sinbad]

Thank you, so basically they just give you the passport to the internet i.e. the access, and aside from that, manage client bandwidth allocations etc.

So then I guess.. even I can start up my own ISP from my house and sell accounts. But that aside, it then means that the problem lies where? Where is this issue happening? It cannot be the residential FTTH connection because if it was that, then it would hold true for a local speed test also

Of course it can be the residential ftth connection. In fact that's the most likely cause.
Packet loss on an ultra low latency connection doesn't have a terrible effect on throughput. Your fibre connection to a local server is very very low latency...
As soon as the latency increases, (overseas connections) the effect of intermittent packet loss becomes significantly more noticeable in a drop in throughout.

 

[wingnut771]

Of course it can be the residential ftth connection. In fact that's the most likely cause.
Packet loss on an ultra low latency connection doesn't have a terrible effect on throughput. Your fibre connection to a local server is very very low latency...
As soon as the latency increases, (overseas connections) the effect of intermittent packet loss becomes significantly more noticeable in a drop in throughout.

How do you fix packet loss?
How long does packet loss take to fix is the better question.
More than 6 months.

 

[wingnut771]

I will wait 7 days then cancel.

 

[Sinbad]

How do you fix packet loss?
How long does packet loss take to fix is the better question.
More than 6 months.

The trick is identifying where it is happening and then getting the owner of that particular piece of the puzzle to acknowledge the problem and fix it. It could be faulty hardware or a faulty splice somewhere or a bad patch lead...

 

[wingnut771]

The trick is identifying where it is happening and then getting the owner of that particular piece of the puzzle to acknowledge the problem and fix it. It could be faulty hardware or a faulty splice somewhere or a bad patch lead...

Here's the problem, what problem, there's no problem.

 

[Wasabee!]

The trick is identifying where it is happening and then getting the owner of that particular piece of the puzzle to acknowledge the problem and fix it. It could be faulty hardware or a faulty splice somewhere or a bad patch lead...

Well there is an easy way to find out if an end user doesn't know how to use or interpret a trace route.

If you don't have local issues ito loss, but have international issues, it's probably ISP. If the ISP tells you otherwise, imo they are lying.

If you have local issues ito loss aswell as have international issues, it could be your fibre provider which your ISP should resolve. A proper ISP has a good relationship with their suppliers and can work with them to resolve faults. The end user shouldn't have to know where the fault is as your commercial relationship is with the ISP.

So they should have some vendor management/relationship and resolve upstream issues whether it is their IPC (internet breakout) supplier or the layer 2 last mile provider like Vuma or Frogfoot.

 

[Sinbad]

Well there is an easy way to find out if an end user doesn't know how to use or interpret a trace route.

If you don't have local issues ito loss, but have international issues, it's probably ISP. If the ISP tells you otherwise, imo they are lying.

If you have local issues ito loss aswell as have international issues, it could be your fibre provider which your ISP should resolve. A proper ISP has a good relationship with their suppliers and can work with them to resolve faults. The end user shouldn't have to know where the fault is as your commercial relationship is with the ISP.

So they should have some vendor management/relationship and resolve upstream issues whether it is their IPC (internet breakout) supplier or the layer 2 last mile provider like Vuma or Frogfoot.

And I'm telling you that say 1% packet loss you will not notice AT ALL on local connections, unless you specifically sniff your traffic for retransmits/window adjustments etc.

I speak from experience. Had crappy international performance on another ISP, great local performance, and fortunately my fibre provider has a local speedtest which doesn't touch the ISP infrastructure. I did speedtests on this, while analysing traffic, and provided tcpdumps showing excessive packet loss despite getting my line speed. The ISP and the provider replaced the NNI and all was good from then.

As to CI's international capacity:


That's not terrible for a 100mb line.

 

[wingnut771]

Well there is an easy way to find out if an end user doesn't know how to use or interpret a trace route.

If you don't have local issues ito loss, but have international issues, it's probably ISP. If the ISP tells you otherwise, imo they are lying.

If you have local issues ito loss aswell as have international issues, it could be your fibre provider which your ISP should resolve. A proper ISP has a good relationship with their suppliers and can work with them to resolve faults. The end user shouldn't have to know where the fault is as your commercial relationship is with the ISP.

So they should have some vendor management/relationship and resolve upstream issues whether it is their IPC (internet breakout) supplier or the layer 2 last mile provider like Vuma or Frogfoot.

You would think.

 

[wingnut771]

And I'm telling you that say 1% packet loss you will not notice AT ALL on local connections, unless you specifically sniff your traffic for retransmits/window adjustments etc.

I speak from experience. Had crappy international performance on another ISP, great local performance, and fortunately my fibre provider has a local speedtest which doesn't touch the ISP infrastructure. I did speedtests on this, while analysing traffic, and provided tcpdumps showing excessive packet loss despite getting my line speed. The ISP and the provider replaced the NNI and all was good from then.

As to CI's international capacity:


That's not terrible for a 100mb line.

Single thread please.

 

[Sinbad]

Single thread is not great, 10mb/sec - but that does not point to a capacity issue. Points to a packet loss issue somewhere.
Then again, I've never gotten good single thread results off the Structured Communications server. Normally I used Vodafone London one but it seems to be seriously borked.

I see vodafone is back online.
Single is around 35. I've seen better in the very recent past. Something's funky tonight.

 

[Tinuva]

@Wasabee!
Not everything you post is 100% accurate.
Even if the OAN only does layer2, if the layer2 is congested (yes it does happen) then it will affect the customer experience. The problem is, with FTTH you don't see this congestion as high latency like you would with ADSL. You _only_ see if when you do speedtests, local will be full line speed, and international will be slow. That is what congestion looks like on FTTH networks. Blaming the ISP in this instance like you are doing is actually wrong. Coolideas is one of the best ISPs to work with the OAN and FTTH providers to get to the bottom of these issues. You should give them a lot more credit than you do.

 

[Wasabee!]

And I'm telling you that say 1% packet loss you will not notice AT ALL on local connections, unless you specifically sniff your traffic for retransmits/window adjustments etc.

I speak from experience. Had crappy international performance on another ISP, great local performance, and fortunately my fibre provider has a local speedtest which doesn't touch the ISP infrastructure. I did speedtests on this, while analysing traffic, and provided tcpdumps showing excessive packet loss despite getting my line speed. The ISP and the provider replaced the NNI and all was good from then.

As to CI's international capacity:


That's not terrible for a 100mb line.

Look I agree with you on the loss be it local or international which is what my last post said. I perhaps didn't say too clearly how people should identify/test loss, but my last post categorised based on where the loss is and where the fault will probably (NB probably) be - as you might've seen in previous posts there is a lot of misunderstanding in blaming the open access networks for international speed issues which is just the ISP passing the buck.

I think you should refrain from using the term NNI as end consumers here won't know what it is or which one it is. NNI is too generic when troubleshooting as it could be an INNI (fibre patch lead) or UNI-C (Ethernet patch lead). Not sure why FTTH users would want to use MEF terminology.

If you say you replaced your fibre patch lead then others will know what you talking about, NNI could be any interconnect or patch lead.

 

[Wasabee!]

@Wasabee!
Not everything you post is 100% accurate.
Even if the OAN only does layer2, if the layer2 is congested (yes it does happen) then it will affect the customer experience. The problem is, with FTTH you don't see this congestion as high latency like you would with ADSL. You _only_ see if when you do speedtests, local will be full line speed, and international will be slow. That is what congestion looks like on FTTH networks. Blaming the ISP in this instance like you are doing is actually wrong. Coolideas is one of the best ISPs to work with the OAN and FTTH providers to get to the bottom of these issues. You should give them a lot more credit than you do.

I agree that an OAN network (or respective ENNI) can be congested. This isn't always the fault of the OAN, as the ISP might not buy sufficient bandwidth on the OAN.

Actually it's very easy to see congestion, you just identify where your loss is. I don't mean to be rude, but the bold text I've highlighted shows me you are not very experienced with networks, again no offence. You don't need to do international speedtests, have a look at programs like WinMTR, MTR if on Linux, Pingplotter etc. They all identify loss which is a symptom of congestion.

Also fibre networks, especially GPON networks in SA due to their relative young age have so much spare capacity. That is on the access network. On distribution and transit on one cable you can have 144 fibres as an example. On each fibre you can multiplex with DWDM another few hundred channels ON EACH FIBRE. Each channel can be 100Gbps or more. So capacity on any good fibre network isn't a problem. Once again it comes down to design and commercials. Just like many ISPs save on international bandwidth, so to do they on OAN bandwidth. So often it might not be the fault that there isn't capacity available, the ISP might not want to buy more local capacity.

I'm not discrediting Cool Ideas, I'm just seeing a lot of people not be aware:

- That you can measure congestion with loss. Forget about slow speeds for now, first identify loss.
- How to identify loss (it seems everyone just likes to run speedtests).
- Based on where the loss is, what the probability of where the issue is (local or international).
- Then run speedtests after you have an idea where along the route the loss might be and use speedtests as another tool to confirm your assumption.

I'm not trying to make anyone look bad, I just want the correct party to take responsibility and not pass the buck by promoting or not correcting incorrect information.

 

[underdose]

My results on same server with CISP 50/50 on Vuma:

Obviously UL is sheit but little to no problem with the DL.
Also, why are we testing a single threaded connection? Can anyone remind me?

 

[wingnut771]

@Wasabee!
Not everything you post is 100% accurate.
Even if the OAN only does layer2, if the layer2 is congested (yes it does happen) then it will affect the customer experience. The problem is, with FTTH you don't see this congestion as high latency like you would with ADSL. You _only_ see if when you do speedtests, local will be full line speed, and international will be slow. That is what congestion looks like on FTTH networks. Blaming the ISP in this instance like you are doing is actually wrong. Coolideas is one of the best ISPs to work with the OAN and FTTH providers to get to the bottom of these issues. You should give them a lot more credit than you do.

After 6 months, nobody deserves any credit.