Cool Ideas Fibre ISP - Feedback

Status
Not open for further replies.
image132 said:
I've tried it before and it made no difference. I get single digit throughput without the proxy every night. My phone usually backs up my pc results.
Click to expand...
if your phone concurs, then go openserve, i wouldnt wait for octotel.
 
wingnut771 said:
if your phone concurs, then go openserve, i wouldnt wait for octotel.
Click to expand...

I mean look at the difference right now.

No proxy:
8201103247.png


Proxy:
8201104642.png


And although 30mb on single for a 100mb line is pretty good for me that will continue to drop throughout the day until peak time when I'm looking at between 3-8mb. Which of course isn't enough to stream twitch on. I still have no idea why the proxy gives me such perfect results.

*EDIT

and just to add some confusion to the mix. Locally if I run single thread tests they're perfect, proving there is no congestion on my OLT card or whatever its called. Its only when my traffic has to leave Cape Town that things go pear shaped.

Cape Town
8201113595.png


Randburg
8201114774.png
 
Last edited:
image132 said:
I mean look at the difference right now.

No proxy:
8201103247.png


Proxy:
8201104642.png


And although 30mb on single for a 100mb line is pretty good for me that will continue to drop throughout the day until peak time when I'm looking at between 3-8mb. Which of course isn't enough to stream twitch on. I still have no idea why the proxy gives me such perfect results.

*EDIT

and just to add some confusion to the mix. Locally if I run single thread tests they're perfect, proving there is no congestion on my OLT card or whatever its called. Its only when my traffic has to leave Cape Town that things go pear shaped.

Cape Town
8201113595.png


Randburg
8201114774.png
Click to expand...
its the whole packet loss over distance.

short distance can recover from loss, long distance cant.

proxy is making you "closer" is how i understand it.
 
image132 said:
I mean look at the difference right now.

No proxy:
8201103247.png


Proxy:
8201104642.png


And although 30mb on single for a 100mb line is pretty good for me that will continue to drop throughout the day until peak time when I'm looking at between 3-8mb. Which of course isn't enough to stream twitch on. I still have no idea why the proxy gives me such perfect results.

*EDIT

and just to add some confusion to the mix. Locally if I run single thread tests they're perfect, proving there is no congestion on my OLT card or whatever its called. Its only when my traffic has to leave Cape Town that things go pear shaped.

Cape Town
8201113595.png


Randburg
8201114774.png
Click to expand...

Have a look at this graph:
TCPProblemGraphLarge.gif

TCP does quite a bit better than this nowadays, but it tries to help visualise what is happening,

so lets say you have 2% packet loss on Octotel, your Cpt speed test will be the bar on 10ms and 2%, so +- 10Mbps (obviously way better these years, you can probably multiply these figures by 10)

Your international speed test will be 2% and 100ms, so give or take 1mbsp, about 10% of your local speed,

Running through the proxy means you can get the good 10ms speeds to the proxy, and the proxy does not need to deal with the packet loss, so the proxy can get the 100ms and 0.1% loss bar, +- 4mbps, 4 times as fast as your speed without the proxy.

And regarding Jhb slowing down already, you can see what a huge difference latency makes going from 10 to 100ms if you have some packet loss, so the extra latency is already going to influence speeds to Jhb.

After all this, this may not be the only problem just to make things super confusing.

We all dream of 0% packet loss. On an idle line, my smokeping tests are about 0.1% to most international locations, and below 0.1 local when everything is happy. I am happy with my line even though there is some room for improvement.
 
image132 said:
I mean look at the difference right now.

No proxy:
8201103247.png


Proxy:
8201104642.png


And although 30mb on single for a 100mb line is pretty good for me that will continue to drop throughout the day until peak time when I'm looking at between 3-8mb. Which of course isn't enough to stream twitch on. I still have no idea why the proxy gives me such perfect results.

*EDIT

and just to add some confusion to the mix. Locally if I run single thread tests they're perfect, proving there is no congestion on my OLT card or whatever its called. Its only when my traffic has to leave Cape Town that things go pear shaped.

Cape Town
8201113595.png


Randburg
8201114774.png
Click to expand...
imho, speedtests dont show the true picture anyway.
8201146076.png

and yet
Untitled.png
 
This is what I get right now during off-peak:

iperf3.exe --port 17001 -c trcvmh01.cisp.co.za --bandwidth 80M -l 1400 --omit 2 -R -u
Connecting to host trcvmh01.cisp.co.za, port 17001
Reverse mode, remote host trcvmh01.cisp.co.za is sending
[ 4] local 192.168.1.247 port 60314 connected to 154.0.15.181 port 17001
[ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams
[ 4] 0.00-1.00 sec 9.74 MBytes 81.7 Mbits/sec 0.053 ms 184/7481 (2.5%) (omitted)
[ 4] 1.00-2.00 sec 9.54 MBytes 80.0 Mbits/sec 0.053 ms 0/7143 (0%) (omitted)
[ 4] 0.00-1.00 sec 9.51 MBytes 79.8 Mbits/sec 0.050 ms 22/7147 (0.31%)
[ 4] 1.00-2.00 sec 9.54 MBytes 80.0 Mbits/sec 0.059 ms 0/7142 (0%)
[ 4] 2.00-3.00 sec 9.54 MBytes 80.0 Mbits/sec 0.058 ms 0/7143 (0%)
[ 4] 3.00-4.00 sec 9.54 MBytes 80.0 Mbits/sec 0.056 ms 0/7145 (0%)
[ 4] 4.00-5.00 sec 9.53 MBytes 80.0 Mbits/sec 0.077 ms 0/7141 (0%)
[ 4] 5.00-6.00 sec 9.53 MBytes 80.0 Mbits/sec 0.065 ms 0/7141 (0%)
[ 4] 6.00-7.00 sec 9.53 MBytes 80.0 Mbits/sec 0.055 ms 0/7141 (0%)
[ 4] 7.00-8.00 sec 9.54 MBytes 80.0 Mbits/sec 0.045 ms 0/7143 (0%)
[ 4] 8.00-9.00 sec 9.48 MBytes 79.5 Mbits/sec 0.065 ms 46/7145 (0.64%)
[ 4] 9.00-10.00 sec 9.53 MBytes 80.0 Mbits/sec 0.074 ms 0/7140 (0%)
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams
[ 4] 0.00-10.00 sec 95.8 MBytes 80.4 Mbits/sec 0.055 ms 68/71441 (0.095%)
[ 4] Sent 71441 datagrams

iperf Done.

Just to make life that little bit better the proxy doesn't always work and almost never on a Sunday.

https://mybroadband.co.za/forum/threads/cool-ideas-fibre-isp-feedback.793061/post-23289882

I'm assuming results like this are caused by my local packet to the proxy right? It just makes no sense.
 
Last edited:
bHOLDher said:
Have a look at this graph:
TCPProblemGraphLarge.gif

TCP does quite a bit better than this nowadays, but it tries to help visualise what is happening,

so lets say you have 2% packet loss on Octotel, your Cpt speed test will be the bar on 10ms and 2%, so +- 10Mbps (obviously way better these years, you can probably multiply these figures by 10)

Your international speed test will be 2% and 100ms, so give or take 1mbsp, about 10% of your local speed,

Running through the proxy means you can get the good 10ms speeds to the proxy, and the proxy does not need to deal with the packet loss, so the proxy can get the 100ms and 0.1% loss bar, +- 4mbps, 4 times as fast as your speed without the proxy.

And regarding Jhb slowing down already, you can see what a huge difference latency makes going from 10 to 100ms if you have some packet loss, so the extra latency is already going to influence speeds to Jhb.

After all this, this may not be the only problem just to make things super confusing.

We all dream of 0% packet loss. On an idle line, my smokeping tests are about 0.1% to most international locations, and below 0.1 local when everything is happy. I am happy with my line even though there is some room for improvement.
Click to expand...

Good post on this issue, but what is puzzling is that @image132 and myself have had similar issues whereby there is almost 0% PL reported to the CPT iperf server even during peak hours (at least in my case) but international speeds fluctuate heavily especially during peak hours. We all seem to be of the understanding that PL is a byproduct of a certain issue such as a noise-heavy line or bad equipment, etc, and not something that changes based on latency (slower throughput would be the result of PL + latency?)

I.e. if one has 0% PL to a CPT iperf server, but 1.5% to JHB and 3% to UK, where does the problem lie? And how is it resolved since the FNO will simply look at the most local iperf and say, nope, no problem with the line.
 
DeatheCore said:
Good post on this issue, but what is puzzling is that @image132 and myself have had similar issues whereby there is almost 0% PL reported to the CPT iperf server even during peak hours (at least in my case) but international speeds fluctuate heavily especially during peak hours. We all seem to be of the understanding that PL is a byproduct of a certain issue such as a noise-heavy line or bad equipment, etc, and not something that changes based on latency (slower throughput would be the result of PL + latency?)

I.e. if one has 0% PL to a CPT iperf server, but 1.5% to JHB and 3% to UK, where does the problem lie? And how is it resolved since the FNO will simply look at the most local iperf and say, nope, no problem with the line.
Click to expand...
keep changing isps
 
DeatheCore said:
Good post on this issue, but what is puzzling is that @image132 and myself have had similar issues whereby there is almost 0% PL reported to the CPT iperf server even during peak hours (at least in my case) but international speeds fluctuate heavily especially during peak hours. We all seem to be of the understanding that PL is a byproduct of a certain issue such as a noise-heavy line or bad equipment, etc, and not something that changes based on latency (slower throughput would be the result of PL + latency?)

I.e. if one has 0% PL to a CPT iperf server, but 1.5% to JHB and 3% to UK, where does the problem lie? And how is it resolved since the FNO will simply look at the most local iperf and say, nope, no problem with the line.
Click to expand...
In the Frogfoot case it is a bit more complicated, if we run a throughput test from a customer and we hit a ceiling without any packetloss. It's almost as if the TCP window just doesn't scale over the Frogfoot lines. We are trying some SDN kit to try give us some more visibility as to why.
 
image132 said:
Just to make life that little bit better the proxy doesn't always work and almost never on a Sunday.
Click to expand...

Everytime you've posted here about the proxy not working, international in general has been slower than usual for me as well on vumatel in cpt. So its safe to assume when the proxy isn't working, that its cisp related.

There is clearly an issue with octotel on your line though, i mean you aren't even getting 100mbps to jhb on single threaded, at least to that server you tested with.
 
Local:

International:


Lovely. Waiting on the router you are delivering, not sure how that's going to help at all considering directly plugging into the ONT doesn't result in better international speeds either.
Frogfoot, TV, CT, 100/100Mbps.
 
PBCool said:
In the Frogfoot case it is a bit more complicated, if we run a throughput test from a customer and we hit a ceiling without any packetloss. It's almost as if the TCP window just doesn't scale over the Frogfoot lines. We are trying some SDN kit to try give us some more visibility as to why.
Click to expand...

And in the case of Octotel are we just waiting for their upgrades to complete? Any idea when that will be?
 
Last edited:
DeatheCore said:
Good post on this issue, but what is puzzling is that @image132 and myself have had similar issues whereby there is almost 0% PL reported to the CPT iperf server even during peak hours (at least in my case) but international speeds fluctuate heavily especially during peak hours. We all seem to be of the understanding that PL is a byproduct of a certain issue such as a noise-heavy line or bad equipment, etc, and not something that changes based on latency (slower throughput would be the result of PL + latency?)

I.e. if one has 0% PL to a CPT iperf server, but 1.5% to JHB and 3% to UK, where does the problem lie? And how is it resolved since the FNO will simply look at the most local iperf and say, nope, no problem with the line.
Click to expand...

Its bizarre to me that we live on opposite sides of Cape Town, with different FNO's and yet we have the same breaks in twitch streams as if we lived next to each other. Something about Cape Town is messed up.
 
Last edited:
Status
Not open for further replies.
Back
Top
Sign up to the MyBroadband newsletter
X