Understanding contention ratio's

Sounds like fun my turn

You have a can of Coke. It costs you R7 for this can. You can sell this can to someone for R7 but they won't necessarily drink it all in one go or you can put two straws in and sell it for R3.50 to two people but if they want to drink at the same time they get less coke but assuming some law that I cant quite think of right now, they will most likely use the straws at different times. According to ICASA you shouldnt have more than 20 straws/ coke but we all know that some ISP's would find their pricing "unsustainable" at these contention ratios so they put more straws in the coke.

Lets move on to throttling vs Shaping

Ok so you have your straws and coke and someone wants to drink(HTTP) but someone else on one of the other straws wants so spit coke at a friend(P2P) the ISP can manage the network so that the drinker gets more coke more quickly than the spitter. Thats shaping ie: only certain things are slowed down

Throttling is when both the drinker and the spitter get much less than their straw can manage no matter what they want to do. Ie: all protocols are shaped.

Why does it have to be Coke? Can't it be Pepsi?
 
Hey Shayd - I have to say that I'm enjoying our conversations on this... one point on the contention issue though,

If IPC, bandwidth and setup & admin costs are R6750 per 1Mbit then it's R27k per 4Mbit - which on a 50:1 base is R540.
The fact that there are some companies who have the local peering agreements and can get it for less per Mbit maybe as low as R4k then you're still talking R320 at the very best case.... and this is wholesale costs before you build in customer support and admin.
Also, we haven't even factored in redundancy/failover for the SEACOM international line, which is just good and professinal business practice.

The money that MWeb have invested in this project is huge and the sheer number of customers that they already have should help them sustain the investment and hopefully get them a peering agreement with one of the big two to significanttly lower their local bandwidth costs (if they havent already).

This, however is why bandwidth prices dropped and only someone like MWeb could have done it. No other Tier 2 in the industry had the numbers or access to the cash and media to pull this off. It is also why other businesses say it isn't sustainable. The idea of spendign R200-300m (for just a 2GB network) just on infrastrucutre is scary, not to mention the huge advertising spend they must have.... however, their shareholders own the South African media... makes life and your advertising budget a little more simple to manage.

Yet they still don't have redunancy.

Companies are now releasing products that are deperately trying to compete on price and name 'uncapped'... I am scared that those will be the only two things focused on for the next few months and quality will go out the window. I'm also scared that the networks and rural Telkom exchanges will not be able to cope with the increase in volume.

There are so many positives to come out of the uncapped issue and sustainability isn't one that I care about. I know it can be sustainable if the economy of scale is in place. however, at what cost to the quality and detriment to the infrastructure as more and more people sign up to a system that is too old to handle it...

I hope we don't end up back in the dark ages because we're too keen to lower prices and Telkom don't have hte hardware to deal with the influx.
 
Last edited:
Unless the backhaul costs drop significantly the only ways you can reduce costs would be higher contention and strictly enforced AUP's or worse customer service. Believe it or not the downloaders may very well find themselves in the dark, 50-75 gigs on 10% of your users is sustainable if they download late at night. The 100's of gigs BS is not going to work in a few months no matter what ISP you are with as they are going to "fire" you off their network or provide you with a severely degraded service. We are already seeing this on Afrihost.

Frankly nobody is going to care as I guess less than 1% of all users use over 75 gigs. When new cables land there is going to be lots of pressure on Telkom to reduce line rental costs as this is now the majority of the cost.
 
Hey Shayd - I have to say that I'm enjoying our conversations on this... one point on the contention issue though,

If IPC, bandwidth and setup & admin costs are R6750 per 1Mbit then it's R27k per 4Mbit - which on a 50:1 base is R540.
The fact that there are some companies who have the local peering agreements and can get it for less per Mbit maybe as low as R4k then you're still talking R320 at the very best case.... and this is wholesale costs before you build in customer support and admin.
Also, we haven't even factored in redundancy/failover for the SEACOM international line, which is just good and professinal business practice.

The money that MWeb have invested in this project is huge and the sheer number of customers that they already have should help them sustain the investment and hopefully get them a peering agreement with one of the big two to significanttly lower their local bandwidth costs (if they havent already).

This, however is why bandwidth prices dropped and only someone like MWeb could have done it. No other Tier 2 in the industry had the numbers or access to the cash and media to pull this off. It is also why other businesses say it isn't sustainable. The idea of spendign R200-300m (for just a 2GB network) just on infrastrucutre is scary, not to mention the huge advertising spend they must have.... however, their shareholders own the South African media... makes life and your advertising budget a little more simple to manage.

Yet they still don't have redunancy.

Companies are now releasing products that are deperately trying to compete on price and name 'uncapped'... I am scared that those will be the only two things focused on for the next few months and quality will go out the window. I'm also scared that the networks and rural Telkom exchanges will not be able to cope with the increase in volume.

There are so many positives to come out of the uncapped issue and sustainability isn't one that I care about. I know it can be sustainable if the economy of scale is in place. however, at what cost to the quality and detriment to the infrastructure as more and more people sign up to a system that is too old to handle it...

I hope we don't end up back in the dark ages because we're too keen to lower prices and Telkom don't have hte hardware to deal with the influx.

Is Imagine having issues right now with their uncapped offerings? (like Afrihost/Mweb) I am looking around for Uncapped now
 
Doubt it they sell straight IS and unless the whole of IS is Pawned they should be fine.
 
Is Imagine having issues right now with their uncapped offerings? (like Afrihost/Mweb) I am looking around for Uncapped now

Yeah - the Afrihost IS product is different from ours.
They control the shaping and we are insisting that the Tier 1 provider does it as they have a huge amount more experience and also capacity/capability. I'm sure they'll sort it out though.
We are keen to add on services to the ADSL connection as a Tier 2 should, providing ADSL should be the 'basic' part of this and as transparent as possible.
 
Frankly nobody is going to care as I guess less than 1% of all users use over 75 gigs. When new cables land there is going to be lots of pressure on Telkom to reduce line rental costs as this is now the majority of the cost.

This would be lovely... if only Telkom wasn't losing money and cash flow at a very steady rate year on year. This likely means that they couldn't affrod to invest in anything without a knock on effect.

I hope that the large infrastructure investment they need to do to bring the country 'up to speed' means they can amortise the money over a significant number of years for it not to effect customer costs. There is almost zero point them doing things that results in even more extortionate charging. Although, it seems that's how Eskom are doing it...
 
A land line telephone has become a premium product and this is where Telkom has screwed up, they know this and their shareholders know this. For Telkom to increase their penetration into the "normal" market again I would do this:

1.) Combine the Line and Adsl rental into one.
2.) Bond a basic ADSL service onto the normal line for free, service at 128k or so excluding cap
3.) Reduce Adsl offerings to : R200 for a 1024k line and R350 for a full speed line.
4.) Allow those who resell the current line to act as agents for the entire Telkom service, eg Imagine IPS now bills me for the line, my data and my telephone calls. This will instantly put an efficient friendly face on a company that is no longer regarded as such and provides ISP's the opportunity to be true convergence providers and eliminates the 2 bills issue with Telkom.
5.) Add a couple of "free" minutes to the standard line so that people don't feel they are getting screwed for a "dumb" pipe.

When it boils down to it Telkom needs to increase penetration, simple line price increases are not going to fly with their shareholders.
 
Couldn't agree more,

However, they still have the monopoly on IPConnect which every business has to buy to supply ADSL.
It's the one thing that separates Telkom from everyone else... they can pay 'cost price'. Assuming the DSLAMs aren't archaic there is a possibility to be very clever and upgrade everyone to xMB lines at one price and keep charing the same price for IPC. This could kill the competition who could not buy the IPC needed to sustain a service at current prices. They could then undercut the market significantly.

It would be a terrible move where a private business, with a heredatary monopoly, could once again control the whole market because of their government links. However this is the position they are in. Essentially it's their choice to either do somethign to help, do somethign commerically violent or do nothing... I wonder what it will be.
 
Well let me shed some light on the subject as no one here has it exactly right. In South Africa the ADSL connections at the DSLAM are contended at 20:1. However there is still a significant amount of contention that gets put on top of that. The norms in South Africa are as follows: 100 users - 10mb pipe 10:1 ratio. 1000 users - 40mb pipe 25:1, 10000 users - 200mb pipe 40:1 and then 80000 users 133:1. These are all real world contention ratios on providers that provide only capped services. If you think that is bad you can easily double these figures for uncapped providers.

Secondly the issue of bandwidth costs it works as follows: you have your per mb price on the IPC, local (SAIX or IS) and international (Seacom or SAT3). You need bandwidth on all three to provide services. Typically depending on the amount of bandwidth you take it will cost between R4000 and R6000 per mb on EACH circuit meaning a cost of anywhere between R8000-R12000 depending on your national/international split. Most companies don't offer redundancy because the costs are insane. If you are an ISP offering Seacom bandwidth you would typically have to pay 70% extra per megabit to have redundancy on a 3:1 basis via SAT3. It would have a significant impact on cost models.

Lastly at the current prices of roughly R30 per gig an ISP needs a lot of users to sustain them. They only start making a profit at those prices when they have either more than 10,000 users or if the average usage per user at,say, 4000 users is 4Gb or more.

Moral of the story - don't even think of starting an ISP if you don't have a very significant competitive advantage because you will shoot yourself in the foot. The minimum OPEX for even a small ISP is about R60k per month and that's without bandwidth or labour costs included.
 
If users hate Telkom, i can only imagine what isp's think.
 
With the IPConnect, where does Telkom do the handover to each ISP network?
What infrastructure is between the DSLAM and each ISP network? (where does each POP (point of presence) come into play?)
How many Tier 1 ISPs do we currently have in SA?
 
With the IPConnect, where does Telkom do the handover to each ISP network?
What infrastructure is between the DSLAM and each ISP network? (where does each POP (point of presence) come into play?)
How many Tier 1 ISPs do we currently have in SA?

Hmmm, I don't think you will get a 'straight' answer to these questions....

However, due to an 'event' on Mon 8th I can give you a partial answer.
Outage Num:9075
MWEB IPC outage affecting Cape Town
MWEB IPC services in the following area may be Slow : Cape Town
Start Date:2010-11-08 13:27:22
End Date:2010-11-08 20:41:52
Status:Closed

According to a response to my complaint, this was caused by a fiber break at the main Barrack St Exchange in CT:
http://mybroadband.co.za/vb/showthread.php/268237-LATENCY!!!?p=5012693&viewfull=1#post5012693

Given that a break in one fiber caused latency to spike to 300-600ms, you can assume that there is a 'narrow' connection between ADSL users and the ISP:
http://www.imagehost.co.za/image-A78B_4CD81EAE.gif

Sorry for the belated reply, but I hope this is of some use to you. Sometimes one learns a lot about the internal workings when something 'breaks'!
 
Please don't take the analogy literally as it has many technical "holes" and doesn't take into account things like caching and shaping. But to get an idea why prices dropped for downloaders, its basically due to those who don't use the net for downloading subsidizing those who do.

Makes alot of Sense
 
What annoys me is some ISPs quote their different 'usage tiers' with arbitrary number sequences. I know the difference is contention rations but there is no way of knowing the actual differences/ratios
 
Top
Sign up to the MyBroadband newsletter
X