Saturday, 29 June 2013

NBN: "Internet speeds will always be lower under a Coalition Government"

This is what the Coalition are actually saying with their "Cheaper, Sooner, More Affordable" line:
Internet speeds will always be lower under a Coalition Government.
It's a little deeper as well: "More Affordable" undeniably translates into "Makes less revenue".
This questions the profitability of the Coalition pretend-NBN: it's not Broadband if its not fast.1
The Coalition will not only offer lower speeds, their pretend-NBN will run at a loss.
A little bon-mot explaining the size of the financial problems of the Coalition pretend-NBN is buried in the Coalition's unverified (more 'pretending') financial forecasts:
They claim the pretend-NBN will be finished sooner and be self-supporting faster, but the whole project accumulated losses are around 50% higher. The gap between Capital Expenditure and Project Funding is the losses incurred while the business builds up. For the Coalition it's $9 billion ($29.5B - $20.5B), $3 billion or 50% higher than the $6 billion ($43.35B - $37.4B, using "Free Cash Flow" & CapEx, 2011 to Dec 2020 from 2012 NBN Co Plan).
 I've not seen anyone tease this out, nor ask "Mr Copper" and friends to explain how this works, or the source of the 50% higher losses by 2020, despite them loudly proclaiming "we're in the Black!".

Somehow, the Coalition and the Member for Antiquated Copper rack up massive losses, but are at a loss to explain where these massive extra costs arise or why those unexplained expenses won't kill the ability of their pretend-NBN to ever make a profit. My rough estimates are that copper DSL/FTTN services can't break-even in 10 years under the current NBN Co charges. With aggressive competition, that translates to never.

There is a very good reason that the Coalition's Misinformation and Dissembling Unit deliberately hid  critical figures of their forecasts:
The way they've structured the finances of their pretend-NBN, it can only make a long-term loss. Not only are the early losses 50% higher, the on-going costs are higher and revenue-growth much, much lower. The Coalition pretend-NBN will be a financial disaster and they know it.
Why else do they stop their spreadsheets at 2020, yet quote totals for longer periods? On the one hand they are saying "we know the forecasts and we've run the numbers", on the other "but we won't show you". I've asked the right people for this data and a clarification, in return I got abuse. This wasn't accidental nor incidental - they are incredibly touchy about it.

Political Parties only hide information about Election Policies when it's bad news. We can infer they are hiding something massive and nasty...

My cynical side says "the Policy you take to the election isn't the one you implement". Does the newly anointed and lauded Inventor of the Internet in the Australia, have a completely different plan he'll announce, along with an O.B.E. for services to UK Telecommunications, 12-18 months after the election.

What's the bet they'll suddenly discover "our pretend-NBN will make a loss and we can't allow that! Better to sell now into private hands, realise a massive loss and crow endlessly about 'Worst Government ever' and 'Big Expensive White Elephant'"?

I am cynical enough about the Coalition to believe they could coldly and deliberately plan to destroy $10 billion or more of taxpayer money just to make a political point while also pursuing an ideological agenda: the privatisation of all profits and the unspoken flipside we saw in 2007/8, the socialisation of corporate losses.

My belief/opinion as a single voter is of no consequence. What matters is: what the electorate thinks on polling day.

1it's not Broadband if its not fast.

In Computer Performance Analysis, there are some very well established principles that have been in use since the 1960's to measure the perceived performance of a system.

It's not internal metrics like CPU % Utilisation or VM activity or paging, but something simple that you can test and confirm with a stop-watch:
User Response Time, measured as statistical 'mean' (average) and proportion of responses longer than 2-3 times the mean.
For transactional systems, mean User Response Time, the wall-clock wait time the user experiences from 'hitting enter' to being able to proceed with the next transaction/step,  is typically 1 second, with no more than 5-10% of transactions longer than 3 seconds.

1 second is special. Below it, most people cannot discriminate between faster or slower responses.
Around 1 second, people at terminals starts to notice the wait, faster than that, it appears instantaneous to them.

With webpages, the usual browser rendered image response time target is 5 seconds, sometimes as low as 2.

Translation: On the web, most people feel waiting up to 5 seconds for an action to complete is "instant".

This metric gives us an on-going way to measure and define "network performance":
Application Response Time not link bandwidth, throughput or latency.
Fast broadband can be defined as "Actions in my application seem to happen without delay".

If you were running a 56kbps modem in 1995, downloading simple HTML pages seemed "pretty zippy". They were small, low-complexity pages that loaded and rendered in the apps and systems of the day, within a few seconds.

Trying to upload an image or video file now over a 1.5Mbps (25 times faster than 1995) will seem glacial because the files are now so much bigger, the wall-clock time taken is excessive - and the user cannot do anything else to distract themselves while they are waiting.

Microsoft, even, was acutely aware of this distraction effect, often explained as "putting mirrors in elevator lobbies", in early versions of Windows. Perhaps in 1995, they introduced background printing with much fanfare and attention, because the system would seem much faster to users. Rather than wait for MS-Word to laboriously render pages of a document ready to print before being able to proceed, sometimes 5-15 minutes, users could get on with editing their documents without waiting for the page-rendering.

So fast Internet was, is and will continue to be, "I don't notice waiting for the next thing to happen".

As the World Changes, so people and countries have to change to keep up

Right now in Australia, we have a very large cadre of ISP customers that are at, or beyond, the mean distance from a telephone exchange, 3km, and since 2002/3 when ADSL2/2+ was standardised for use in Australia, have been locked into 3-5Mbps connections or even 1.5Mbps.

While those link speeds were fast a decade ago, they are now slow and limiting, because the world has changed, but not the link speed available.

The most notable example of "the world changed around us" was the Y2K bug.
All the systems, databases and software that worked perfectly up until 31-Dec-1999 and then would've failed, weren't suddenly changed. All their code was identical to what it had been, but was now wrong.

The software and systems stayed unchanged, but the world changed around them: as the calendar and clock ticked over to 01-01-2000 00:00:00, all that code became faulty, not before.

This is the nature of "bit-rot" in Computing and I.T.: the world changes and our assumptions, buried in the code and configuration, don't.

What is "fast" Broadband today will not be considered "fast" in 10 years time: We can guarantee that.

That's the central problem in selecting a suitable technology for a mass rollout in every country. Not "what can we afford now" nor "what do we need today", but
"what will be barely adequate when we throw this new network away?"
It's seems perverse, but Telecomms Engineers must begin their calculations for new parts of the Network with forecasting its demise and replacement. This is the central distinction between DSL/FTTN and GPON/FTTP: VDSL will start mostly adequate and be generally inadequate well within a decade.

Some Network Changes are Invisible, Getting Fast Broadband to the Premise is NOT

The trite statements of the Opposition, and Members for Never Changing Anything, that
"Networks are always being upgraded"
are highly misleading, even, deceptive. A Fibre Customer Access Network is a replacement, not an upgrade project. You cannot do that cheaply, quickly or without disruption.

Those parts of the Network that can be replaced without customers noticing, do get upgraded as demand increases, because they were designed that way. Inside the network, there is never, ever a "Single Point of Failure". There are always 2-3, sometimes even hundreds, of different paths that data can follow.

The system is designed for continuous operation with automatic, transparent failover, so your phone call or internet surfing can continue uninterrupted no matter what disasters are befalling the equipment and links you can't see.

The largest part of the Network, "the last mile" or Customer Access Network (CAN), isn't duplicated.
For every ordinary residential and business customer, there is one massive Single Point of Failure for all their services: the Customer Access Network.

We saw this in spades with the Warrnambool Local Exchange fire. Local Exchanges are where the Customer Access Network joins the highly reliable trunk network: they are still a Single Point of Failure for subscribers.

What is wrong and misleading about the statement "Networks are always being upgraded" is leaving out, "but only in the parts that were designed to be upgraded." In early 2009, the Government Expert Panel made the observation that DSL/FTTN was a dead-end, investing in one had now become a road to nowhere. The Coalition Minister for Yesteryear wants to pretend it's 1999 again and we should party, celebrating lower-speeds over ultimately more expensive copper.

VDSL/FTTN becomes a worse idea every year

The Customer Access Network was never designed to be upgraded to Optical Fibre. It's a replacement, not an upgrade.

The fudged solution that is DSL and FTTN rely on the Copper Customer Access Network alone and cannot be upgraded without massive cost, disruption and effort (equals time).

Blithely saying, "Networks are upgraded in place all the time", while ignoring the critical and significant engineering and financial/economic factors surrounding that, is at very best, uninformed or disingenuous.

A VDSL/FTTN network is, today, a ticket to nowhere.

In 2000-2003, ten years ago, it was a valid temporary network solution while a brand-new Customer Access Network of high-spec Copper (such as TransACT) or pure Optical Fibre was built. As each year has passed since then, the speed and cost equation has shifted more solidly away from Copper to Optical Fibre. It's not so much on the input, or cost, side but on the output side: Customer now want and need much more speed, reliability and durability than degraded Copper can offer, and the gap is only going to increase.

That is why in the early 1990's Frank Blount, then Telstra CEO, planned, and is on record as saying:
[attributed quote: "1995 Strategy Paper". updated 15-feb-2015]
'we expect to have replaced the (entire) Copper Customer Access Network by 2010'.
That is also why in 2005 Sol Trujillo, just 6 weeks into his term, went to Canberra and addressed Howard personally, telling him the Copper network had to be upgraded and replaced, immediately.

The Coalition sat on it hands then when it could've acted, again in 2007 with an inconsequential Broadband policy, and again in 2010 and now for the 2013 election, they proffer another bad deal: exhorting us to "save 10% in return for a 40 times slower network that we'll throw away in 10 years!"

No comments:

Post a Comment

Note: only a member of this blog may post a comment.