Charting Fibre speed over time.

[jacof]

You are testing a local server in the same metro, <100Km vs gfmalan is testing from Cape Town to Johannesburg where the route is over 1500Km, these are very different conditions. And so not relative at all, change his to best latency as well and graph both connections.

Paul we did do those test as well. Then you told me via email and I still have the mail that the test are not valid because it is not against your test server on your network. So we changed the code to force the test to your server. Now you come up with another reason why it is not valid after you stated we had to do it that way. If there is latency it will be on your network then as everything in the chain is CI other than the last mile carrier when gfmalan does his tests. I can change the code again to find the best server closest to gfmalan. It will be the same results I saw when I used the best server closest to me testing on the CI network.

 

[jacof]

I'm going step in here before a fight breaks out since I was the one who asked jaco for the vox test.

The last graph for me wasn't about testing Gfmalan's line vs Jaco's vox line. It was looking at how stable Jaco's line was on VOX vs his line on the old graph. Assuming Jaco tested his line to the same server in both graphs the graph is valid. At least for my purposes. All I wanted to see was the change in Jaco's line.

@image132 I don't want to fight. I am not a CI client at this stage. I found the ISP that works the best for me at this point. Who knows if that will change as I was with MWEB for 17 year before the wheels came off. I just reported what I saw. I would suggest that people do their own test if possible and make up their own minds.

 

[SpiderGear]

@image132 I don't want to fight. I am not a CI client at this stage. I found the ISP that works the best for me at this point. Who knows if that will change as I was with MWEB for 17 year before the wheels came off. I just reported what I saw. I would suggest that people do their own test if possible and make up their own minds.

So your connection might be more "stable" but Vox is going to throttle your line. In my opinion you went backwards in terms of service.

 

[PBCool]

Paul we did do those test as well. Then you told me via email and I still have the mail that the test are not valid because it is not against your test server on your network. So we changed the code to force the test to your server. Now you come up with another reason why it is not valid after you stated we had to do it that way. If there is latency it will be on your network then as everything in the chain is CI other than the last mile carrier when gfmalan does his tests. I can change the code again to find the best server closest to gfmalan. It will be the same results I saw when I used the best server closest to me testing on the CI network.

No you are confused, the first tests via IPC I asked you to do so, this is now a separate thing altogether. It wont be the same results as it is a different operator (we saw how this varied with people on Octotel vs Vumatel on the same ISP). You are again comparing a test via national backhaul vs to a local server which introduces latency to the equation, this has a major impact on the way that TCP operates so you would get far more skewed results regardless of the network.

 

[jacof]

So your connection might be more "stable" but Vox is going to throttle your line. In my opinion you went backwards in terms of service.

I have not seen that. I have downloaded large ISO images via torrent sites and got 12MB/s. The fastest downloads I ever got to date.
They state the do not shape or throttle, if they do then I will report them to ICASA for false advertising of a service.

 

[jacof]

No you are confused, the first tests via IPC I asked you to do so, this is now a separate thing altogether. It wont be the same results as it is a different operator (we saw how this varied with people on Octotel vs Vumatel on the same ISP). You are again comparing a test via national backhaul vs to a local server which introduces latency to the equation, this has a major impact on the way that TCP operates so you would get far more skewed results regardless of the network.

Setup a test server in Cape Town and gfmalan can test against it.

 

[PBCool]

Setup a test server in Cape Town and gfmalan can test against it.

He can use best ping same as your graph.

 

[jacof]

He can use best ping same as your graph.

OK we changed the code. It will test that way now.

 

[PBCool]

OK we changed the code. It will test that way now.

Cool how often does it poll, as all I see is other traffic on that line at the moment.

 

[gfmalan]

Cool how often does it poll, as all I see is other traffic on that line at the moment.

Every 10minutes

 

[PBCool]

Every 10minutes

Cool I see it now.

 

[gfmalan]

Cool I see it now.

Also locked it to use Clear Access server everytime.

 

[PBCool]

Also locked it to use Clear Access server everytime.

Yeah I saw that as well.

 

[jacof]

Also locked it to use Clear Access server everytime.

Seems that your upload is faster but your download is still up and down a lot but above 80Mbps most of the time.

 

[gfmalan]

I think I'm done with testing. (Unless someone want me to test something)

Because I'm in Cape Town, the closest CISP SpeedTest server for me is Randburg (the only), that gave me ok results but not great. Avg 26ms delay, and the graph for the first bit, then I changed it to test against a Cape Town based server (Clear Access).

I know this is a FTTH product, and if it means I have to loose 20% of my speed during peak times, then I'm happy with that. To get a FTTB solution would cost me 10 times this.
To loose 90% isn't on, and that is why I started with this, but since have been resolved with Tickets to CISP and Octotel and comms with PBCool.

So inshort, I'm happy with the service, and I'm not moving anywhere.

The fact that PBCool interact with us makes the difference for me, I know how it is to battle with IS and other ISPs to get something working.

And ping results to few servers:

First server is in JHB
Second in the UK
Third Cape Town
Last, maybe Germany

 

[Samstel]

Hi, been reading this thread and thought i might take a chance and ask for some help, a bit of info first.

I have been with Crystalweb on Openserve since June 2017 i noticed that every night like clockwork 6:30pm i start to see in game latency spikes (League of Legends) usually lasts till 10pm. I have logged calls with them on which they responded all looks well after testing etc, eventually after taking multiple screenshots i got them to log a call with IS, and also explained that my 3rd hop on a tracert which points to afrinic and has average ms of 40-50ms and is in Sandton, is not "acceptable", but not sure if this is the only problem or maybe its congestion on my network, as i live in a complex with roughly 200 units so it could be this PON or GPON everyone keeps mentioning, so here is my question.

Is there anyone who is willing to share details to a non IS account that i can test, before i cancel and switch and cause more frustration to myself, if have tested with a work colleague's account unfortunately it was also IS.

Hope someone can help me out, because if i have to die one more time to latency i might just break my very pretty keyboard and desk

 

[Seeyou]

Hi, been reading this thread and thought i might take a chance and ask for some help, a bit of info first.

I have been with Crystalweb on Openserve since June 2017 i noticed that every night like clockwork 6:30pm i start to see in game latency spikes (League of Legends) usually lasts till 10pm. I have logged calls with them on which they responded all looks well after testing etc, eventually after taking multiple screenshots i got them to log a call with IS, and also explained that my 3rd hop on a tracert which points to afrinic and has average ms of 40-50ms and is in Sandton, is not "acceptable", but not sure if this is the only problem or maybe its congestion on my network, as i live in a complex with roughly 200 units so it could be this PON or GPON everyone keeps mentioning, so here is my question.

Is there anyone who is willing to share details to a non IS account that i can test, before i cancel and switch and cause more frustration to myself, if have tested with a work colleague's account unfortunately it was also IS.

Hope someone can help me out, because if i have to die one more time to latency i might just break my very pretty keyboard and desk

I believe both Afrihost and Webafrica have free trial accounts that you can use to compare, though Webafrica is also on the IS backbone.

The entire Crystalweb feedback thread in the ADSL ISP section is filled with people complaining about latency in the evenings though, so you are not alone.

 

[Samstel]

Unfortunately Afrihost don't do it anymore so no luck there

 

[jacof]

Hi, been reading this thread and thought i might take a chance and ask for some help, a bit of info first.

I have been with Crystalweb on Openserve since June 2017 i noticed that every night like clockwork 6:30pm i start to see in game latency spikes (League of Legends) usually lasts till 10pm. I have logged calls with them on which they responded all looks well after testing etc, eventually after taking multiple screenshots i got them to log a call with IS, and also explained that my 3rd hop on a tracert which points to afrinic and has average ms of 40-50ms and is in Sandton, is not "acceptable", but not sure if this is the only problem or maybe its congestion on my network, as i live in a complex with roughly 200 units so it could be this PON or GPON everyone keeps mentioning, so here is my question.

Is there anyone who is willing to share details to a non IS account that i can test, before i cancel and switch and cause more frustration to myself, if have tested with a work colleague's account unfortunately it was also IS.

Hope someone can help me out, because if i have to die one more time to latency i might just break my very pretty keyboard and desk

I will ask cavedog if we can use his test account for you to test with. You can also get an Openserve test account for a day to check.

 

[Samstel]

I will ask cavedog if we can use his test account for you to test with. You can also get an Openserve test account for a day to check.

You will so be my hero i just tried to get a 100GB fat pipe adsl from vox out of desperation, but the payment was trolling me. Hopefully you can help me out im just soo fed up with laggggggg, its driving me nuts i stop moving in game for easily 3-5 seconds. Sigh