Fastest fibre ISP in South Africa

elp

Active Member
Joined
Dec 11, 2013
Messages
70
Something fishy going on here.

That's because its rubbish.

Speed test servers must be the most optimised unshaped connections in the country. For everything else the average ISP shapes the traffic aggressively. Even 'unshaped' should really be sold as "less shaped". My 50/5 fibre package usually reports around 52/4.95 on speed test but it's rare that I exceed 20M even when downloading something local.
 

Skerminkel

Expert Member
Joined
Dec 3, 2008
Messages
3,741
Are average speeds really a useful metric for FTTx?

Most users select a package for the best balance between speed, data and price. Most therefore probably do not select the fastest speed that the ISP may provide. This means that the large consumer ISPs (Webafrica, Mweb, Afrihost,...) will have many clients who are satisfied with a lower speed. Average speeds for them will therefore be significantly lower.

A test of how close speed tests are to the line limit would be a better indication of quality of service.
 

Johnatan56

Honorary Master
Joined
Aug 23, 2013
Messages
29,976
Are average speeds really a useful metric for FTTx?

Most users select a package for the best balance between speed, data and price. Most therefore probably do not select the fastest speed that the ISP may provide. This means that the large consumer ISPs (Webafrica, Mweb, Afrihost,...) will have many clients who are satisfied with a lower speed. Average speeds for them will therefore be significantly lower.

A test of how close speed tests are to the line limit would be a better indication of quality of service.

+1
Ask the person doing the speedtest what the promised speed was, compare as a percentage of promised package achieved.
ISP gets a nice 95% or in the case of Telkom, usually 90% due to copper limitations.
 

Rickster

EVGA Fanatic
Joined
Jul 31, 2012
Messages
19,294
That's because its rubbish.

Speed test servers must be the most optimised unshaped connections in the country. For everything else the average ISP shapes the traffic aggressively. Even 'unshaped' should really be sold as "less shaped". My 50/5 fibre package usually reports around 52/4.95 on speed test but it's rare that I exceed 20M even when downloading something local.

Is Fibre a "best effort service", I dont think so, complain to the ASASA.
 

Mr Scratch

Expert Member
Joined
May 15, 2013
Messages
4,819
But its fibre? There shouldnt be any factors that affect the speed like copper does.

If there is no SLA, there is NO expectation of service beyond what is reasonable from the advertisement. The type of service does not matter (telecoms, logistics, food etc) and this does not only apply to lil Africa, it applies to industries all over the world, from all different nations.
 

MickeyD

RIP
Joined
Oct 4, 2010
Messages
139,117
Anyone using a speed test article to choose their service provider deserves all the pain that they will suffer later.
 

TheRoDent

Cool Ideas Rep
Joined
Aug 6, 2003
Messages
5,713
These results are a problem, since obviously the average download statistics are affected by the type of package that the ISP's customers use on average.

For example, Bitco and iConnect probably have a large number of corporate customers with 500mbps + packages, whereas the others have more consumer customers with packages varying between 40-80mbps.

Thus the more consumer oriented ISP's will never rate at the top, on average speeds. However are their networks any slower ? I think not.
 

MickeyD

RIP
Joined
Oct 4, 2010
Messages
139,117
These results are a problem, since obviously the average download statistics are affected by the type of package that the ISP's customers use on average.

For example, Bitco and iConnect probably have a large number of corporate customers with 500mbps + packages, whereas the others have more consumer customers with packages varying between 40-80mbps.

Thus the more consumer oriented ISP's will never rate at the top, on average speeds. However are their networks any slower ? I think not.
Spot on. And how do you eliminate the ISPs own testing of their own Gig or 10Gig, etc. links?
 

Skerminkel

Expert Member
Joined
Dec 3, 2008
Messages
3,741
Is there a system which can accurately test this?

You are asking the wrong guy. I am only here to criticize! :D
Seriously though, I really don't know. Your article just made me think about the utility of averages. I assume to test the metrics I suggested would be much more effort = expensive. Maybe as the biggest and most effective broadband lobby in the country you could investigate?
 

koeksGHT

Dealer
Joined
Aug 5, 2011
Messages
11,858
Spot on. And how do you eliminate the ISPs own testing of their own Gig or 10Gig, etc. links?

^

I could automate 20k tests on spare gig link to get an average speed of 1G.

You can tell the differences as you see Cool Ideas/Home ISP's will average under 100Mbps as that's residential speeds, but now you are comparing to companies that sell corporate links like Seacom/Bitco. Not to mention the muckery which looks like self testing to keep up averages :whistling:
 
Top