Monday, September 13th, 2010 by Simon Hackett
Fairly comparing broadband ADSL plans in Australia can be difficult
In this post, we take a look at some common differences between provider plans and think about the impact of each of them on making a valid comparison.
This is a long post, and if you aren’t a ‘detail person’, this is probably not a great post for you to start reading – because you probably won’t finish it :)
(and ironically, thats kinda the punchline here too)
The default question that seems to be at issue for most consumers when first approaching broadband ADSL plans in Australia is ‘how many gigabytes do I get for how many dollars a month?’
This deceptively simple question shields a multitude of real world differences – some of which appear on the pricing pages of ISPs concerned and some of which do not.
When looking at this list, have a think about where your personal preferences land around this old saying:
Good, Fast, Cheap: Pick Any Two.
(Any pair of those tends to work. Expecting all three at once… generally doesn’t work)
First, lets start with the non-numeric things that matter:
Now lets look at the measurement approaches that are usually at issue when doing a ‘desk comparison’ of plans, and get a sense of how tricky it can be at times to compare Apples to Apples:
Now, lets look at two specific items in the list above, that deserve special attention.
1) Upload counting
In the last year or two, there has been a steady trend toward ISPs moving to offer new plans where the monthly quota is the sum of the data uploaded and downloaded, rather than the more historically popular tendency to count downloads alone.
This is a vexed issue for any ISP with a history of download-only quota assessment, because the “Apples to Apples” comparison between one of their plans and the plan of a provider offering “upload+download” plans is impossible to do in a manner that applies equally and fairly.
As quotas have started to expand dramatically over time, this places both commercial and fair-comparison pressures on an ISP to move to upload+download counting on those newer, larger plans.
To address these two points separately:
Commercial (cost) pressures:
The higher customer upload speeds become, the greater the real contribution to cost that occurs from data uploads.
When the highest possible speeds for Australian ADSL services were 1500 kilobits per second (downstream) and only 256 kilobits per second (upstream), this meant that the peak network load possible in the upload direction was relatively limited.
The ADSL2+ era resulted in upstream speeds in the 1 megabit range.
The emergence of ADSL2+ Annex M from a few innovative providers has increased the peak upstream speed to rates of up to 2.5 Megabits per second (sometimes even higher, on very good lines). That rate is faster than the maximum possible downstream data rate only 5-6 years ago.
The National Broadband Network is already offering routine speeds in the 100 Megabit realm, with peak speeds already framed to hit 1 Gigabit per second.
As more symmetric applications become popular amongst consumers (peer to peer videoconferencing, peer to peer content exchange, etc), these large changes in network utilisation pattern also raise per-user average cost for an ISP over time, all other things being equal.
These two trends (increases in peak upstream data rate and the evolution of network usage patterns) have a multiplicative impact on the total upstream data rates carried in modern Internet networks.
As a result of this, the real cost contribution of uploaded data to the operating costs of Internet Service Providers need to be factored fully into network operating cost modelling.
Network capacity for an ISP must be expanded when a link hits peak capacity in either direction. Many years ago, the download direction was the dominant source of network peak utilisation events. These days, at various times of the day, either the ‘download’ or the ‘upload’ direction can be the dominant network peak demand source.
Router and switchgear costs for an ISP are also a major cost, and the capacity of this equipment is a function of the sum of uploads and downloads – because routers run out of horsepower when they hit a total data throughput and/or total packets-per-second limitation, regardless of which ‘way’ the packets are flowing through the device.
In summary, the more consumer usage trends head toward symmetric network usage, and the more routinely consumers use much higher upstream speed connections, the more the real cost to the ISP has to be factored fairly into the commercial constructs used for broadband plans.
Comparison fairness (Apples to Apples) pressures:
Its a reality of the modern world that so much of our decisions are made with nothing more than a quick sound byte to base them upon.
When comparing two ISP propositions, the tendency is for the process to become as simple as this:
Quick: This ISP offers x GB’s per month for $50, how much does that other ISP offer for the same price? Which offers more?
In that quick comparison, no accounting is made of whether one ISP counts uploads and the other doesn’t (and little credence to peak vs off-peak either, but we’ll come to that).
The difference in effective quota is quite hard to predict in general. That is because it depends on whether your own usage patterns place a high emphasis on uploaded data or not.
Its clear that the answer to this question doesn’t just vary per person, either – it also varies over time for each of us, as our requirements and habits evolve.
Some users have a ratio of uploads to downloads that can be of the order of 1:8 or even less, others can approach or even exceed unity (1:1). It all … depends.
For quota, then, these two drivers are additively placing pressure upon newer/better/stronger/bigger ISP plans to count uploads, in order to be fairly compared on a quota size basis to other existing ISPs (including the largest in the country) who count uploads routinely and who have done so for a long time.
In essence, if everyone does this, then the comparisons between ISP quotas at then at least fair and and reasonable to make.
It is not surprising, then, that almost every major ISP now does this in Australia on at least some of their plans (and in many cases, on all of them).
2) Time-Of-Use quota restrictions
This is a more vexed issue in many ways than ‘upload counting’ is.
Its an interesting instance in which the answer to the question “is 5+5 equal to 10” is “it depends”.
To make this more interesting, lets look at a desk comparison between a 150GB ‘anytime’ quota from one ISP and a 100GB (peak) + 100GB (offpeak) quota from another one.
Is 150 more or less than 200?
It depends on whether, and to what extent, you actually use that ‘off peak’ quota.
Its made even more tricky by the related question of how much you care about offpeak quota, and whether you actually ‘needed’ more quota at off peak times, or whether its just a question of psychology (of having it there ‘just in case’ – despite almost never using it).
By its nature, an off-peak provider makes their off-peak quota available for only a minority of the time. That might be as much as 11-12 hours, or as little as 3-4 hours in each 24 hour cycle.
For the sake of some examples, lets consider an off-peak quota that applies for 8 hours per night (e.g. from 1am to 9am in each 24 hour cycle).
This means, by definition, that you have access to that off-peak quota for 8/24 or one third of the time in the month (at best), and noting that these are the hours of each day in which most of us are fast asleep for most of it.
Unless you are transferring content 24×7 across your whole month (which almost never applies in the real world), its now a question of whether you are prepared to modify your Internet usage habits to suit your ISP, to optimise that quota usage, or whether you just don’t want to have to care that much about your ISP telling you how you have to change your life to suit their network
Of course, for the technically minded, its eminently possible to use a download scheduler, to arrange for specific content downloads to start at the commencement of your ISPs off-peak period and to have them shut down automatically at the end of that period.
If you actually have such an organised life and such personal requirements in how you life your life, such that you have a stacked up list of content to access at all times, you can therefore pile them up carefully in advance each night of the month and keep the ‘pipe’ full every night.
That is in huge contrast to someone who uses the Internet when they want to use it (‘on demand’), and who is likely to prefer having just one quota for Internet access.
The larger your off peak quota allowance, and the lower your line speed, the more likely it also becomes in the real world that you just can’t use the full quota efficiently without going to increasingly ‘extended’ measures to do so, literally every night of the month.
And again, for some people, this is all just too hard – and they’d prefer to access the Internet when they want to, not when their ISP says they can.
While generalisations are, as I’ve explained, really hard to make, it seems clear that an offpeak quota is worth less than an on-peak one, because of the access time limitations that apply, combined with an (obviously) higher potential that you won’t be awake to use the off-peak period and won’t have cared to arrange download schedulers to be ticking away downloading… stuff… to fill in those overnight hours for the sake of it.
One rule of thumb to consider here is to account for an off-peak quota as being ‘worth’ around 1/3 to 1/5 of the stated value (midpoint around 1/4), when these issues of time of day, line speed, and time coordination are factored in.
So that would suggest a 100+100 quota might be reasonably compared to a 120-133 GB ‘anytime’ quota, for most people.
I will say, again, that its clear this comparison and its validity depends on each person concerned, and their preparedness (and desire) to go to great lengths to optimise and consume every last byte of quota each month.
As a result, for some people, “100+100” is really “100” because their computers sleep when they do… and for others, its “200” because they super-optimise their use of the plan concerned.
Here is another way to look at the off-peak question, just to explain that whatever your view, an ‘anytime’ quota is different to a peak+offpeak one.
Imagine you had an ISP with a 1000 gigabyte quota, expressed as 500GB (on peak) and 500GB (off peak), where you have 8 hours a night to use the off peak quota.
What if that was 100 GB (on peak) and 900 GB (off peak). Would you consider that to be the same ‘value’ at 500+500?
What if it was 50 GB (on peak) and 950 GB (off peak). Is that the same as 500+500?
And: Do you need an ‘anytime’ quota, by comparison, to be ‘1000’ in order to unconditionally equate to any of these?
Or do your computers sleep when you do, so that an off peak quota that can only be used after 1am really doesn’t mean that much to you at all?
In summary, lets look at three more quick comparisons:
a) 100 Anytime vs 100 Peak +100 Offpeak … Winner: Peak/offpeak (it can’t possibly be worse than 100 anytime, even if you don’t use any of the offpeak quota)
b) 150 Anytime vs 100 Peak +100 Offpeak … Winner: Its debatable (depends on your time of day and real world usage requirements)
c) 100 Anytime vs 50 Peak +50 Offpeak … Winner: Anytime (because it even a perfectly utilised peak/offpeak cycle can’t possibly exceed the arithmetic total offered as a single anytime value)
Simple numerical comparisons can be confounded, in this realm, by differences in measurement technique, and by differences in time-of-day banded usage limitations, combined with differences in consumer usage habits. The comparison can also vary over time as consumer habits evolve.
Comparisons between ISP plans in Australia are often unfairly or incorrectly made, because consumers don’t care about all these nitty gritty details.
This is what does tend, over time, to drive the emergence of similarities in quota presentation and assessment across ISPs.
Like it or not, similarity in the underlying measurement approach yields fairer ‘instant’ comparisons for consumers.
Meantime, to borrow (and mess with) an old advertising catchphrase:
Quotas aint Quotas.