PlusNet and Central Capacity

The following information started off as a summary to try and understand the situation with PlusNet and Central Capacity Issues during 2006. This report has not been sanctioned by PlusNet and it is intended to be unbiased and as factual as possible based around information to hand*.

NOTE: This data is now significantly out of date - but the principles remain the same. For this reason PUG has left this page available as a guide for others to update using more recent data.

The figures are complicated and we have tried to simplify them where possible, in the hope that they may enlighten users to some of the realities of ADSL within the UK. Please do not quote any information contained herein out of context as this document has to be read in its entirety to understand what is happening.

For those of you on other ISP’s, look at the figures at the bottom of this page and ask how is any other IPStream ISP going to be able to cope that much differently?

Overall Picture

*It’s very important to note that the following information and stats are NOT official PlusNet figures. The vast majority of information has come from sources already in the public domain. On occasion figures have had to be estimated, and in these cases an explanation is given as to how this estimate has been achieved.

Feb 06	These figures are estimated as per explanation below.
Speed Mb	Users %	No users 160,000	Bandwidth Users	Capacity 2635 Mb
				Segments
0.5	10	16,000	8,000	17
1	20	32,000	32,000	Sessions
2	69.78	111,648	223,296	100,880
4	0.02	32	128	68%
8	0.20	320	2,560	Ratio
				99.92
All	100	160,000	263,296	100:1

Based on end Jan 2006

Feb 05	% breakdown obtained from information posted on AG Feb 2005
Speed Mb	Users %	No users 100,000	Bandwidth Users	Capacity 2635 Mb
				Segments
0.5	65.98	65,980	32,990	17
1	24.33	24,330	24,330	Sessions
2	9.69	9,690	19,380	124,800
4	-			125%
8	-			Ratio
				29.11
All	100	100,000	76,700	29:1

Based on end of Jan 2005 - These figures were calculated Feb 2005

Explanation of these figures.

Customers - 160,000

In reality the figures may be slightly higher than this. The figure errs on the side of caution - we know PlusNet reached 150,000 ADSL users in November 2005.

The % of customers on each particular bandwidth

Over the past year, the BT Wholesale (BTw) bulk regrade process has been in place. Ian Wild said last year that BTw anticipated that 90% of customers should be able to get 2Mb. However, since then BTw have tightened up the rules as to which lines can get 2Mb. Also judging from what has been seen on the forums, there was/are still a few that haven’t been upgraded.  
At the current time, therefore it is quite realistic to assume that the % of customers that were on 512kb last year, is near the figure now on 2Mb. As a result, it is felt that the 2Mb and 512 kb % are likely to have simply "switched", and the % of customers that were previously on 2Mb is now likely to be nearer the true % of customers still on 512kbps.
The % of customers on 8Mb has been based on a figure stated in the Portal forums that were on the maxDSL trials (over 300 users).
A small percentage of these trialists will not have been able to get the full 8Mb - probably less than 10%.  
However, when maxDSL is fully implemented, these percentages will obviously change again, and customers across all bands should be able to get much higher speeds.

Contention Ratio

Today’s market place sees a much higher proportion of lower-type-usage customers. We are led to believe that the ISP industry anticipates 1/3rd of the userbase being online at any one time sharing available bandwidth. Therefore specifying a fair level of service makes more sense than a contention ratio (which was difficult for many users to understand).
In Nov 05 BTw quietly removed the word "contention" from their product descriptions. Instead it has been replaced with acceptable speeds for a contended product. PUG asked PlusNet in our November meeting to mirror this move when describing their own products and we are hopeful that this will happen.

Overall

Obviously the percentage of customers on each particular speed band are rough estimates, but from observations it would appear that these are not too far from the mark. I’m happy to be corrected if they are more than a couple of % out either way.

Increase In Growth

Customer Growth

The past year has seen phenomenal growth.
In the past year the customer base has risen from 100,000 to 160,000 (60% growth). Bandwidth capacity for these customers has increased by 15.65%, or 22.45% over the fiscal year.

In the later part of 2004 PlusNet was having to light practically 1 segment each month, which took capacity costs over budget and spiraling out of control. More info

These figures do not take into account the additional 15,000+ users since the acquisition of Metronet, nor the capacity for those customers (extra segment lit 03/01/06).

Face Value

Growth Feb 05 to Feb 06
Growth	Customers Increase	Bandwidth Requirements	Sessions Increase	Capacity Increase	Adjusted Increase
Increase	60,000	186,596	-1600	0	356.5
%	60	243	-	0	15.65

The true situation

The above figures don’t show the full picture:

Replacing the 10 x 155’s with 2 x 622 pipes after making an allowance for the 23% ATM overhead would have given them an immediate increase in throughput of 46.5Mb. 2 further 155 segments have also been lit during the period Feb 2005 - Feb 2006. Making an overall increase of 356.5 Mb (15.65%).

There was also another segment that was lit in late Jan 2005 which more or less coincided with PlusNet reaching the 100,000 customer mark. This segment can’t be included within the Feb 05 to Feb 06 period, but would obviously still be within the PlusNet fiscal year. This makes an increase in capacity of 511 Mb (22.45%) over the fiscal year.

There is also one very important difference - today’s newcomers to the Internet are very different from the type of user that was chomping at the bit 3 years ago to get ADSL. The early adopters of ADSL were more likely to be "net savvy" and know how to get the most from the Internet.

The vast majority of new sign ups are low usage - this is borne out by the fact that 90% of BB+ users use less than 1.5GB per month [source 1.24 GB]. There is less profit in these accounts and therefore less need for extra capacity on the Centrals.

During mid Feb 2005, PlusNet replaced the 155 Centrals with 2 x 622’s. The 622’s are 23% more efficient on overheads than the 155’s. There will have been large costs involved for PlusNet to transfer over to all 622’s. The cost of 2 Junipers @ approx £500,000 each would have cost in the region of £1m alone.

Important differences between the 155 & 622 Centrals
Pipes	Segments	LTSs	ATM overhead	Sessions
155 Mb	1	2	23%	8,000
622 Mb	4	12	0	25,600

Calculating the adjustment for conversion between 155 and 622 pipes

155Mb v 622Mb additional ATM overhead calculation (23%)
Pipes	155 segments	Capacity	Overhead	Adjusted
155	10	1550	356.5	1,193.5
622	7	1085	0	1,085.0
Total	17	2635		2,278.5

Replaced with 15 x 155 segments = 2325 Mb.
2325 - 2278.5 = 46.5Mb throughput increase.

Sessions

Transfer to all 622’s meant fewer available sessions - "BT Central Extra" should have provided increased session limits but despite BT saying this was on the horizon, it never actually introduced. Unfortunately this caught a number of ISPs out who had designed their products around it.

Although there are 160,000 users, PlusNet only had 108,800 available sessions. That meant that at any one time, only 68% of the customer base could be connected, hence the implementation of Idle Timeouts on the BB+ account.

With effect from early 2006 BTw have implemented a pilot to increase sessions to 32,000 per 622Mb central which in our opinion gives a much more acceptable 85% of the user base being able to connect at any one time, and is much more in line with available sessions on the 155Mb pipes.

Session Limit Calculation
Date	Pipes		Session Limits		Available
Feb 05	10	7	8000	6400	124800
Jan 06	17		6400		108800
Feb 06	17		8000		136000

108800 / 160000 = 68%
136000 / 160000 = 85%

Growth with respect to the Products

We do not have a specific breakdown of the number of users on each type of account. However, since the introduction of BB+ in April 2005 it seems that the vast majority of new users are choosing BB+. It wouldn’t surprise me if about 40,000 of the new customers have chosen this product. There will also be many users who have made the switch from Premier/PAYG to the BB+ product. If asked to make a guesstimate it would be as high as 75-80% of new customers are choosing BB+.
It’s therefore fair to assume that approx one third of customers are on Broadband Plus, with this % increasing as more and more users get ADSL for the first time. [source 47,606 BB+ users in Dec 05 against a customer base of 150,000 = 32%].

Over the past year, it would appear that PlusNet have concentrated their focus on Home users rather than Business customers. Therefore growth on Business accounts (which have more scope for profit in comparison to the home products) has remained low/static in comparison to other products.
Based on the above info this would perhaps leave approx 35-40% of users on Premier with the balance being made up of PAYG and legacy accounts.

Looking towards the future, more first time users will choose BB+, seeing a decline in the overall percentage on Premier, but conversely some BB+ users will move up to Premier or even PAYG as they become more accustomed to what the internet can offer.

General Comments on throughput

Throughput on the centrals

Based on 160,000 customers with an average of 5.28GB (both up and downstream) each across all products would mean a total throughput of 844,800 GB per month.

Capacity

An ISP using CBC has to pay for central capacity whether or not it is used.
Under perfect conditions the maximum capacity on a 622Mbps central is calculated as follows:-

Downstream: 622 / 8 (Bits) * 60 (secs) * 60 (mins) * 24 (hours) * 30 (days) / 1000 (GB) = 201,528 GB
Upstream: 311 / 8 (Bits) * 60 (secs) * 60 (mins) * 24 (hours) * 30 (days) / 1000 (GB) = 100,764 GB

PlusNet Available Downstream: 201,528 / 4 (segments) * 17 = 856,494 / 160,000 (customers) = 5.35GB per customer
PlusNet Available Upstream: 100,764 /4 (segments) * 17 = 428,247 / 160,000 (customers) = 2.68 GB per customer = 8.03 GB total

Guaranteed Clean Usage

This leads on to the guaranteed "clean figure". Having seen that PN’s pipes under ideal conditions allows for only 8GB per person, it’s doubtful that they can really guarantee that each and every Premier customer can actually get 20GB during peak. This will obviously depend on the actual percentage of customers on Premier and how many of those Premier users use less than the average figure

Should they light more pipes?

Surely they should light more pipes?

This is the obvious answer. It’s something that a lot of users on the forums (including some PUG members) have been asking for a long while.
So now for some more calculations... (see below) and it becomes clear that PlusNet can’t/aren’t ever going to be able to do this in the way it was for the period 2004/2005.

Why’s that?

PlusNet still have 2 unlit segments, which contractually they have to light by Feb 2006.
They are also soon to move some customers out to LLU, (a small trial is already underway) therefore it doesn’t make sense to tie themselves up with BTw on a 2 year contract.
(Although Pipe 1 has just come out of the minimum 2 year contract period).

One thing that has been puzzling though, is why they haven’t lit those remaining segments before now?
Yes of course the most obvious answer is that it would cost them an extra £1M pa.
However, after crunching some figures there’s another side of this.

If you look at the figures I’ve compiled for BTw costs alone, these work out at around £165.08 per user per annum - that’s £13.76 per month for each customer going to BTw before they even start on anything else.
The lighting of those 2 extra segments immediately means that the BB+ account gets them zilch profit before you even start on staffing/other costs etc.
With one third of customers on the BB+ product, it certainly doesn’t give much room for manoeuvre.
Therefore when things are on a tight budget it makes sense to delay lighting each central and save yourself about £37,500 each month.

Feb 06	BT Wholesale Costs pa
	segment	ex-vat	segment	inc-vat
Centrals	373560.00	6,350,520.00	£438,933.00	7,461,861.00
Ports	£100.80	16,128,000.00	£118.44	18,950,400.00
to BTw		22,478,520.00		26,412,261.00
£ per user		140.49		165.08

That’s 13.76 per month per user on BTw costs alone.!!!!!!!!!!!!

Now let’s take the maximum throughput figure at note [1] whereby we can see that this £13.76 buys each and every customer a guaranteed 8.03 GB of total throughput (of which 5.35GB is downstream) before any other costs incurred for equipment and staffing.

Feb 05	BT Wholesale Costs pa
	segment	ex-vat	segment	inc-vat
155’s	316,200.00	3,162,000.00	£371,535.00	3,715,350.00
622’s	373,560.00	2,614,920.00	£438,933.00	3,072,531.00
Centrals		5,776,920.00		6,787,881.00
Ports	£100.80	10,080,000.00	£118.44	11,844,000.00
to BTw		15,856,920.00		18,631,881.00
£ per user		158.57		186.32

Note how during the previous year the costs per customer were higher. This was before the implementation of faster speed upgrades and lower monthly costs.

IPStream BT Wholesale Costs

What is clear and has emerged from the above BTw costs, is that for any IPStream ISP to be able to guarantee all of their customers 5.35 GB of downstream usage on fully utilised pipes each customer would have to pay in the region of £13.76 pm on BTw costs alone.

Port cost are £9.87pm so the 5.35 GB of downstream bandwidth per customer is £3.89.

One other thing sometimes asked on the forums is how much more would it cost if I wanted to say download a 5GB movie?
So lets try and work this out:-
We know that £13.76 of which 9.87 is port costs therefore 3.89 buys 5.35 GB of download.

Therefore £3.89 / 8 (maximum throughput) = 49p total GB used or
or £3.89 / 5.35 = 73p per each GB downloaded.

The table below gives an indication of how much the BTw costs alone are to guarantee each user xGB of data throughput each month.
The figures are based on the cost for a 622Mb Central inc VAT.
Don’t forget that ISP’s will have a lot more costs on top of this such as equipment, staffing and other transit costs.

GB to guarantee each customer: Downstream Only     Both Directions Min Monthly cost to BTw to guarantee each customer xGB GB down GB total Port cost Central Bandwidth Cost per month                                                                       This table dynamically calculates the minimum monthly cost to an ISP inc VAT. If an ISP is under utilising Central Pipes this cost will increase. This table is meant as a guideline base cost only. More info how it works

 

Conclusion

Although we can’t be certain of some figures, key members of PUG feel that these figures make a fair representation of the facts to the best of our knowledge.

The cost to BTw per user should be as close as possible to calculate the average. There will be some accounts that cost more such as the Office (old 20:1) products - but those customers pay more each month anyhow for the product. These costs would be the same for any IPStream ISP and are not exclusive to PlusNet.

Without knowing a breakdown of what % of customers are on which product, it’s impossible to calculate any figures that would indicate where surplus profit is coming from and how much capital is spare to fund any additional bandwidth.

The introduction of a £14.99 product will have brought more customers in, but the profit margins on these accounts are low. This product appeals to the low-usage customer, but because of this low profit margin, it means that users on these accounts can’t in any way subsidise those users that require more bandwidth each month.

Allowing customers on this product to download too much which could be the financial detriment to any ISP. IMHO pricing on this level can only sustain a couple of GB at most per month and should therefore be marketed as such. Taking into account the deferment costs of activation and hardware, this particular product could perhaps be seen as a loss leader?

Instead it is probably those users on other accounts such as Premier doing less than 10GB per month that are subsidising those that use more than 10GB per month. Many BB+ account holders don’t use anywhere near the 5GB figure, [source] but those that do, soon eat into the "break-even point" for the product.

Concerns have already been expressed as to how they are actually physically managing to guarantee a clean 20GB per customer, if they get too many users over the 20GB, then because of the traffic shaping it will be to the detriment of those than use less :(
PlusNet have recently issued a statement in response to questions about capacity on the PlusNet network, a copy of which can be found here.

Its not just PlusNet that will find themselves in this situation either. Unless the ISP has funds from other sources to subsidise those needing more bandwidth, then those ISP’s will also be in exactly the same position.

PlusNet perhaps "went wrong" by seemingly trying to camouflage the problems, putting a spin on a lot of news and hyping things up without communicating anything. During this time, other pertinent information was also communicated but often lost within the rest. By not making the real reason behind this clear from the beginning, and trying and wrap it up in "Marketing Hype" clouded the issues and hid the real facts. There are too many users on the forums that are astute enough to see through it and want truth. But then again there are other ISP’s out there guilty of the same and much more so.

The current market is ripe for new comers to ADSL who wouldn’t have a clue and are therefore scared of the terms of caps/FUP/SUPs for different reasons than "the heavies". They wouldn’t have a clue where to start working out costs, nor to spend hours putting something like this together - and why should they have to?

The decreasing of costs to the end user has meant many more users can get broadband. At the end of the day IPStream ADSL has always been designed to run as a contended product. Without the sharing of bandwidth then the true cost would be one heck of a lot more :(

It seems like the days of users being able to download all and sundry are drawing to a close, and those that assume they can because they think they "are paying for it" need to perhaps think again. There is no way any IPStream ISP can sustain too many heavy users without funding from other sources. Until BTw reduce the price of their Centrals then there is nothing much going to change. The only other alternative is an increase in prices across the board and then the low usage users (who are in the majority) would feel hard done by.

The advent of higher 8Mb speeds draws near, and we cant help but wonder how this will affect the end user experience. BTw prices are exactly the same for these higher speeds, so they are perhaps reliant on extra revenue from the ISP Centrals. This in turn makes it harder for ISP’s - it’s obvious that a few users will use more bandwidth simply because they can get things faster. The wholesale costs remain the same to the ISP, therefore they can’t provide this additional bandwidth without either imposing exactly the same "caps" as on the lower speeds, or charging the customer more for the product. Many astute users are also very well aware of what impact these higher speeds may have on exchange congestion.

So is this ADSL of the future? Have we had it good for too long?
Well it certainly seems like it :/ - the early adopters of ADSL certainly did, but then again the costs were so much more and the speeds a lot less than today... and BTw had a different charging method. BTw’s charging method for the Central Pipes is considered by many to be overpriced and perhaps out of alignment with many other EU countries.

Before we start blaming BTw too much for this situation there is a lot more in it. Firstly the fact that BTw have managed to make ADSL available to the vast majority of the population. There is also another side to this story too, in that BTw can’t reduce their prices right now which is down to something called the "Margin Squeeze Test" which has been enforced by OFCOM.
OFCOM has basically ruled that BTw cant reduce their prices to below rates that would give them an unfair advantage over any other telecommunication company. There are already many good articles that have been written about the MST - a good place to start would be AdslGuide if you wish to know more.

So is LLU going to be a savior? That is debatable - more users are adapting to their higher speeds, and will they be in a similar position to what BTw is today? There will come a point when the LLU "centrals" start to fill. LLU could result in a two-tier adsl system in the UK, as they aren’t made to enter into the less profitable exchanges. There are many exchanges and users that will be left out in the cold.

Note by the Author
This "report" is not just something that has been thrown together in a few hours, it’s something that I’ve been mulling over for quite a while, and has taken months to accumulate some of the data. The whole exercise for me has been enlightening in a strange way. - I started off thinking PlusNet were "tight" for not lighting new segments as and when required, but now I’ve had to face a reality check. :/
I don’t like traffic shaping - It’s always something I’ve been against - I still don’t like it and I never will, but the alternative of letting unmanaged contention kick in as and when is even worse.
- I’d much prefer a clean cap or pay a few extra £ per month... I’m also very aware that there is no way I could afford a 1:1 connection.

 

© Kitz - Blame me if I’ve got any of the figures wrong-
Its not my intention to mislead and information stated is to the best of my knowledge.