Broadband Ratings Report: First Draft

[ic]

Just wondering what happened to latency comparisons?

I'm especially preoccupied with latency lately, so I would really like to know how each service scored on latency - perhaps the GRAPHICAL RESULTS SUMMARY could include a latency graph?

 

[bwana]

I agree Gaz{M} - I can just see Telkom misquoting the report to their advantage.

In addition I dont think that saying that the price for any of the local offerings can ever be said to be 'good' regardless of it is comparatively good or not. Were there any other nations that scored worse than SA internationally? If not then this should be pointed out.

Just my 2 bits worth.

BTW - dont let my comments detract from the fact that it’s an excellent bit of work!

 

[martin]

I'm especially preoccupied with latency lately, so I would really like to know how each service scored on latency - perhaps the GRAPHICAL RESULTS SUMMARY could include a latency graph?

Yip, if you have that info available, I second that. I think this will please the gamers no end.

 

[rebel]

> dont say DSL384 if affordable. It is NOT affordable by 99% of South Africans. Say it is the "least expensive offering in relation to overall performance". This clearly states that although it is a good offering, it is still overpriced.
> dont say DSL384 is a "great alternative" to ISDN or dialup. Rather, it is a "potentially cost saving alternative". The general idea is not to give ANYONE the impression that adsl is in any way cheap or affordable.
> For the international section. Mention EXPLICITELY that the percentage scores should not in ANY WAY be compared to the local percentage ratings. I can see it now

""Telkom quoted the MyADSL broadband report, saying: Our HomeDSL384 scored 74% while international offerings such as BT Yahoo 2Mbit DSL only scored 56%, making Telkom's offering significantly better overall."" <--- THIS CANNOT HAPPEN!!!!

Good points.

 

[jasweb]

File is damaged....

Hi RPM

Hope it is not just me... but the file seems to be corrupted...

Can you check please?

JASWEB
----Why does it have to be CAPITAL?----

 

[rpm]

Hi Jasweb

It seem to be just you It opens just fine on my side…anybody else having problems?

 

[stoke]

DSL 512 : that it also carries a 3 usage limit. Missing the "GB".

You could graph based on cost / cap and cost / reliability and cost / downspeed and cost / upspeed.

Also - next month methinks a table at the end detailing all the combinations must be inserted.
e.g.

- Teklom 512 | Telkom ISP Unshapped 4GB | Rn.nn | 510 KB up | 212 dn
- Teklom 512 | Teklom ISP Shaped 3GB | R n.nn | nnn KB up | nnn dn
- Teklom 384 | Teklom ISP Shaped 3GB | R n.nn | nnn KB up | nnn dn
....

 

[rabbiddog]

rpm and co

great work.
jasweb must be you

 

[Gatecrasher]

I tried to send some results for the tests, but the BBST test just didn't work at all with Iburst - there was definitely a serious problem iro latency that caused massive packet errors - not actual errors, but errors in so far as the specific testing software was concerned. Perhaps because the software is designed for first world countries with low latency, the 600ms cutoff is way too low.

I can max out my Iburst connection at 1Mbits/sec at will at any time on multithreaded downloads (local or international). None of the tests use multithreading - yet nearly all download managers use multithreading. Even single threading, I never get below 900Kbps on the the Telkom speed test, yet this local test was not part of the test data. Why not?

So IMO the Iburst evaluation in your report it is totally meaningless, which is rather sad.

I totally agree as to the poor service from WBS, although they would argue that they have not officially launched the product yet. But the performance stats are just ridiculously absurd. How representative are your samples? How many testers did you get? And how many of the submitted tests actually worked?

On another topic. I also have a problem in ranking Vodacom3g as the cheapest broadband option. That is a clear misrepresentation, and dangerous if people accept your report at face value. I think a lot more thought has to go into what is cheap and what is not, because the offerings are so different.

Otherwise, good work!

 

[rpm]

Hi Gatecrasher

Thanks for the feedback!

Let me respond to a few of your points:

The packet loss results played a very small roll (very low weight) in our final result calculation. The main reason was the unreliability of results.

The BBST was only one of three results that were used to calculate the final results (also web based tests and TCPIQ). I think we can not expect testers to conduct more tests than that, and when one considers that the Australian Government based all their findings on only the TCPIQ results I think we can consider these results respectable at least. We will certainly consider local tests the next time if we can find an independent local testing method.

Regarding price: our formula takes everything into account, including installation costs, minimum monthly costs, price per Gig, cost of additional bandwidth, cap sizes etc. Vodacom 3G has the lowest basic monthly cost, but the final percentage gives an indication of the effect of its high BW charges.

As you might know we have also done a user survey to see whether our test results and user feedback corresponds, and it did indeed! iBurst users were particularly unhappy with the service and experienced severe downtimes, lack of service and slow speeds. I therefore think the final result is justified.

BTW: Do you think iBurst should have scored higher? Do you feel it is currently better than V3G or MW?

Regards,

RPM

 

[Gatecrasher]

Was there a questionaire with the testing? Because I didn't get one. Its just that I see you mention a customer satisfaction rating in the report...

On reflection, I suppose you are in a difficult position when compiling a report like this because it requires volunteers, yet volunteers make notoriously bad samples.

 

[Gatecrasher]

BTW: Do you think iBurst should have scored higher? Do you feel it is currently better than V3G or MW?

I can only speak for my own experience. So the answer to both questions would definitely be "yes".

Out of interest, how many Iburst users were included?

 

[rpm]

Was there a questionaire with the testing? Because I didn't get one. Its just that I see you mention a customer satisfaction rating in the report...

On reflection, I suppose you are in a difficult position when compiling a report like this because it requires volunteers, yet volunteers make notoriously bad samples.

There was indeed a questionnaire…no idea why you did not receive one. Customers gave feedback regarding various issues, including customer satisfaction. Vodacom3G scored very high in all departments regarding helpdesk, support & general satisfaction with the service.

I agree that a more suitable sample is possible, but it is highly unlikely that we will be given access to a user DB from all four companies to randomly select users. And then to contact them and where they will then be tech savvy enough to install the software, schedule testing, mail results and perform web based tests is even more unlikely . It is also not suitable to ask the providers for stats for obvious reasons. These volunteer based tests are commonly accepted as a good method to evaluate true performance of the various offerings, and I think the results are quite accurate. Since the average results from the regular testers corresponded with each (both between individual testers and between BBST and TCPIQ) other it further creates trust in the results…

Regards,

RPM

 

[rpm]

I can only speak for my own experience. So the answer to both questions would definitely be "yes".

Out of interest, how many Iburst users were included?

There were 8 iBurst testers that was involved in the ‘full testing’, and a few more that only did one part.

Just out of interest, why do you feel iBurst is better?

 

[Vox Populi Vox Dei]

RPM, although there were seemingly enough iBurst testers, surely it would be unfair to test them before their commercial launch? In terms of different pricing packages, reliability etc?

 

[rpm]

Hi Dominic Rooney

Maybe correct, but I think users are certainly interested. I think it will look great for them if they can show a significant improvement after their launch...

Regards,

RPM

 

[Gatecrasher]

Just out of interest, why do you feel iBurst is better?

This is a personal view.

I guess its because I regularly acheive the 1024/384 speeds. I've had only 1 serious incident of downtime (+-9hrs overnight) in 3 months. I don't need an ISP or Telkom phone line. Overall, its been more reliable than my old ISDN64 line, and much cheaper to run. It also works very well on my home network.

Of course, I will grant that not every iburster has had the same expereince (although many have). A lot depends on the quality of the signal and which base station you are connected to. Northcliffe and Bryanston have been notoriously poor.

I think there is almost certainly a greater variation in the experiences of Ibursters than with say ADSL, that's why it is probably more important to have a representative sample.

If you plug in your modem and get a great connection, Iburst is unbeatable. If you don't, you may be in for a torrid time. But that is why there is a trial period available, so you can test your signal.

So while you may give them a low rating, I suspect that WBS can barely keep up with demand.

 

[ic]

RPM, although there were seemingly enough iBurst testers, surely it would be unfair to test them before their commercial launch? In terms of different pricing packages, reliability etc?

Actually I think "iBurst" should read "iBurst Pre-Launch" or "iBurst Soft-Launch" - those are both phrases used officially by WBS, I think that would be fair.

PS: I have never seen WBS officially using beta testing phase so I would not recommend using the word beta.

 

[Vox Populi Vox Dei]

Oh yeah, what about 'Uncapped' offerings from UUNET and the like? Although it does use the Telscum ADSL service, it would be of interest to me and alot of others probably to know which service [if I wanted uncapped ADSL] is the best.

 

[diabolus]

Looking at the scores laid out like this:

MY Wireless 256______7 - 4 - 7 - 6 = 62%
Vodacom 3G_________8 - 3 - 9 - 8 = 52%

i assume speed carries a -really- heavy weight? Just looks rather peculiar from this angle.