Original URL: http://www.theregister.co.uk/2008/02/22/isps_broke_mailbag/

The amazing, imploding ISP business

Not much sympathy for the tubes operators...

By Andrew Orlowski

Posted in Letters, 22nd February 2008 16:04 GMT

Andrew's Mailbag This week I described how amazingly vulnerable much of the British ISP business is. Vulnerable to botnets, hackers or fraudsters? No, nothing so exotic. A few people watching a bit of BBC on the streaming iPlayer may be enough to bring much of the business here to its knees.

(I drew on a post on STL Partners Telco 2.0 blog, who used data published by Plus.net on the first month's iPlayer - do check them both out).

Some of you might be experiencing a bit of deja vu, here. ISPs rang the alarm some time ago about the impact of iPlayer. We explored the costs of delivering high-definition video over the internet last year here. But this crisis is being precipitated by the low bandwidth, streaming version of iPlayer.

Streamed TV over the internet costs your ISP a penny per viewer per minute. They're obliged to carry it, and there's no additional revenue for them. Which sounds like a mug's game. So what's the answer? STL suggests that the BBC should bear the cost of streaming - but I asked for yours.

A few of you took up the challenge.

First, a dishonourable mention to those that didn't.

I don't know who'll pay (what) exactly, but as a supply system: wifi and swarm?

David

Sorry, David - that's a Fail.

Instead of pumping it into the black hole that is the £40 billion Olympic debt, they could use some of that money to clean up soot covered London and improve the transport system (by bringing it into this century and widening seats for all passengers who are not stick thin pygmies).

Or - here's another idea. Let Google do what it want's to do and pay for a huge cloud wi-max network so that we no longer need to dig up roads and install optical fibre connected at exchanges by copper wire.

Michael G

Sorry Michael, that's corporate charity, and not really a recipe for future investment.

Really simple – ISPs charge based on usage, ‘x’ gigs a month for £|$x.xx with additional usage, charged at the same rate, added onto next month’s bill – might have some impact on piracy as well since it’s a lot of data you’re spaffing about when you P2P a movie. This is how a lot of web hosting companies already charge; the only difference is that they are serving their users data (to everyone else) whilst ISPs are retrieving that data for their users.

So long as we don’t get ripped off on the initial pricing structure (‘shyeah, right’) this could actually work out better for _most_ people – no more ridiculous “fair usage policies” so even extremely heavy users can download whatever they like (as long as they’re prepared to pay for it) and everyone else might see a slight reduction in their bills since they’re no longer subsidising the heaviest users or a small improvement in latency.

Chris Cheale

I think it's heading that way, Chris. It's odd to think that "unmetered broadband" and flat rate access was so hard fought over the years, yet seems unsustainable now.

Brian Griffin is more explicit:

It means that ISPs will have to offer realistic packages that they can afford to uphold. The ISPs will have to narrow their subsidising model and offer more customised packages. No longer will the unlimited package be sold to everyone, based on the fact that 90% of users will use less than 5GB, and only 3% of users will use over 50GB. Instead, the high users will need to pay a premium for an unlimited package, and the majojrity of users should pay less for a lower usage. It should stop some ISPs cramming too many people onto pipes as their customers will start to notice the slower speeds now that the acerage Joe is downloading/streaming video.

David Tomlinson hasn't forgotten the unmetered campaign at all:

There have been similar stories in the past (so recent, so soon forgotten). Brutal Economics and BT have appeared in the same sentence in the past; (OK, so they didn't use the exact words).

Remember the campaign for unmetered telecommunications (RIP 2001 - http://www.unmetered.org.uk/), Why metering is wrong ?

David also takes issue with my assertion that "BT, because it operates an end to end service, can also lower costs a little. But not much."

Not according to the experience of the rest of the world with deployment of optical fibre and WDM (GPON and Active Ethernet). Of course copper and the cable architecture are limiting factors. In Korea 100Mb/s (optical) 30,600 won, costs less than 4Mb/s (ADSL) 31,020 won, about 30 USD.

He continues:

Some developments that may have escaped your attention:

1. Moores Law: When applied to networking equipment we have seen a regular doubling of bandwidth every year, of course if your kit is not state of the art you could be several generations behind, and if you are still using copper wire you could still be in the bronze age.]

I'll jump in here. It's not really a technology problem - so doubling bandwidth is neither here nor there. If the operator can't make money, they won't get capital, and the networks won't get built.

2. BT's 21 Century Network, C21.

By 2012 (sooner in many cases) BT should have a backbone capable supporting the 24Mb/s it is proposing deploying to users (ADSL2+) from the exchange.

3. As noted above the cost per bit of bandwidth falls as the capacity rises, so Japan offers 100Mb/s for the same price as 10Mb/s ADSL in Australia (quote above). Of course it is not just lower cost per bit, but as BT have indicated in their FAQ, deploying fibre lowers the cost of maintenance of the whole network.

Telcordia has a report that states this more than covers the cost of deployment in less than a decade (eight years).

Reduced costs, higher bandwidth, that is the reality of FTTH.

David F

No.3 is cited on David "Net Neutrality" Isenberg's blog however, so the source is more evangelical than credible. The citation is also four years old. As we can see from the stagnant deployments since then, reduced op-ex is not compelling enough.

Looks like we're back to square one.


I can see no problems here, the ISP industry chose to capture customers by competing on price - just on price - now they must change; if they're charged by the gigabyte they should hand in on, as really they always should have.

The only issue I can see is the industries misuse of the language; "unlimited" means what it says as does "free". There is a word for using these words in any other way, it is "lying". This approach has sadly conditioned customers to believe in Santa, the Tooth Fairy and an eternal free lunch. Nearly all the blame here is with the ISPs and, I suspect, their evil wing; the marketing departments where one would hope the P45s would land first (unlikely).

In the end I am more than happy to be metered, especially if it helps in network investments. Those that aren't happy about it are what's called in the real (capitalist) world "customers not worth having".

Steve Hardy

So is it IP TV... or bust?

I can’t agree with you. We have successfully implemented IP-TV schemes for telco operators and it really works. The thing is, the telco operator gets the money for tv. As for watching tv using licit content downloaded through the internet, it isn’t good for the operator but still not so bad, as the user would buy the most expensive package.

A 3Mbit tv stream can be considered “acceptable”. This is roughly 1,3 GB per hour. For 3 hours a day, it would mean 117 GB a month per user. This is something like 48 KB/s, or something like 330 Kbit/s SUSTAINED (remember: you can always throttle) If you sell ADSL2+ 20 Mbits at 25:1 contention rate, you should be Ok. If not, then you just lie when you sell those ADSL lines. And remember: this is 3 hours a day of mostly payperview content. This is the maximum target!!

"A"

Overall, there's precious little sympathy for the ISP industry, from wholesale to retail.

P2P has been with us for years. The fact that the ISPs failed to change with the times only illustrates poor management - they had plenty of opportunity to cap/charge upstream traffic just like downstream, or, like telephone companies charge a profitable rate for downstream. In the race to hit rock bottom prices though they threw survivability out of the window. Frankly I think we'll see what we see during every market crunch - consolidation of the bigger players. Reselling was never a long term business model, as the guy whose goods or services you are selling on will always have a cost advantage. When the crunch comes as it is now, you can bet your supplier looks after himself before he worries about you. The medium sized players like Sky, Tiscali, Carphone Warehouse etc have seen it coming and are pushing LLU as hard as they can to avoid BTs price increases. Everyone else is cannon fodder.

Ross B


Funny how ISPs complain it will all go up in flames.

However, I would like to point out, this is mostly the fault of the very ISP who complain they will soon go burning down. Failure to invest in fatter pipes and severe over-saling have put the ISPs where they are now. Promising to deliver 20MB/sec when you cannot provide it to everyone may gain custumors in the short term, but those users will want to, you know, USE the service you sold them.

Let's use the legendary pipe analogy. If the water compagny says it will deliver 400 litres of water an hour but fail to account for the fact that you may actually use it, then who is to blame when you decide it's time to fill your pool and the water runs out?

If ISPs had concentrated more on delivering on promises made instead of false advertising of "unlimited Ultra Fast" internet, that they then need to cap with "fair usage" policies that, honestly, none of the users who will have been interested in the "unlimited" and "ultra fast" portions of the offer will fall into, then I think it's time regulators took a hard look and said "Alright, start being honest with people and then you can complain".

Paying 60$ a month for internet I can't use as I like doesn't sound like such a good deal to me.

Chaz

I deliberately shied away from discussing the non-appearance of multicast, as a couple of you pointed out. I'll have a look at that next time. ®