Has Telkom fooled the ratings test?

[kaspaas]

Hi,

I'm using Telkom Business ADSL with Telkom Prolog+

I kept the bandwidth testing I did for the report running at 3 tests per day, and sometimes a manual test to shift the time I do the test.

It seems as if the speeds have dropped quite a bit since the tests were officially completed

See http://www.tcpiq.com/tcpIQ/LineSpeed/Results/TestStats/?CountryId=196&ConnectionId=10540&UserId=146951

for the results.

Did the same happen to anybody else?

Regards

[antowan]

Interesting question.

Anyone?

[rpm]

Hi Kaspaas

The graph does seem to change significantly after the 11th of January…

[antowan]

This is very disturbing if true. We know Telkom reads the forum, but to actually affect the outcome by temporarily boosting performance and then letting it slip. That just sucks like a Zimbabwe mozzie! I think we need to investigate this or do continuous testing to keep service levels up...

:)

Cheers
Ant

[lewstherin]

I think a monthly revision of the ratings should be the way forward...with the ratings each month based purely on the previous month's data.

[rpm]

Hi LewisTherin

I agree with that this will be the best option, but that will mean that the testers will have to agree to do continual testing… We will see how the testing goes next month and get feedback from all the testers.

Regards,

RPM

[kaspaas]

Two or 3 tests a day does not munch that much bandwith. I guess there will be quite a natural time distribution.

[stoke]

Yip - like the internet, this monster will change and shift almost daily.
I think that Teklom would not be bothered to try something like that.
It would take far too much organisation.

[beyers]

I think a monthly revision of the ratings should be the way forward...with the ratings each month based purely on the previous month's data.


Maybe true, but difficult in practice. Regular tests eats a lot of bandwith. Hence we may not have that many willing testers each month due to the SA capping culture. And the data processing is also quite a job to repeat monthly ...

[alchamy]

The best way to catch Telkom out is to arrange 2 or 3 testers for a one week period. Use pm instead of the forum. If the results are indeed much lower than during the test phase then it is a valid question.

[ic]

Telkom (as well as other service providers) could theoretically skew things by giving higher priority to the hosts listed in LSM's TestResources.txt - even if LSM knows when a transparent proxy is in place during uploads & downloads. I admit that I don't know enough about what LSM was doing behind the scenes - I speculate that it randomly chose something in TestResources.txt - no doubt making it more difficult for Telkom to skew things. However, Telkom could just intercept LSM traffic & build up a list of international host sites used during testing & proxy the whole lot of them.

<suggested_solution version="1.0">
Application:

Combine the features of a bandwidth meter/monitor with a speed testing application like LSM.
.
100% Removal of Transparent Proxy:

It is really very simple, make sure that the file being sent is encrypted (apply sender's private key, then recipient's public key). The file itself should be compressed data - doesn't matter what, just as long as it has a virtually negligable compression potential (this avoids compression during transmission). You could also place a "sending timestamp" in the file prior to encrypting it. Random filename - perhaps a GUID.</suggested_solution>

[antowan]

Good day,

Again an interesting theory and one that may well be in place. Perhaps there can be a local testing facility and an organised overseas server for testing which can detect lost tests which are affected by diversion tactics on the part of Telkom. It all does seem a little cloak and dagger though, doesn't it? If the nature of the beast dictates, then we will have to be smarter. Telkom did however still lose out on the International scores and if it is true that they skewed the results through some kind of trickery then they might well have the ability to fake a better performance on the International testing, right?

I am personally not too concerned as I believe our strength for forcing better service and better pricing lies with the international comparisons which we now have thanks to the sterling work done by RPM and BEYERS together with the members who so willingly helped with the testing by allowing the testing software to run on their computers.

The only problem here is the possibility that Telkom might be doing us in (worse than usual) at this very moment!

Cheers
Ant

[mikkelz_za]

This is very sad, if true. :/