This article was originally published on 9 November 2010 on newmatilda.com.

With a $43 billion price tag, it’s no surprise that there’s a lot of debate surrounding the proposed rollout of the National Broadband Network (NBN). There’s also a lot of confusion about what the proposed network can deliver, and, in any event, whether it’s worth the projected cost.

Before we begin, let’s outline exactly what the NBN is, and how it’s different from other broadband technologies—acknowledging at the same time that the information available about the NBN is still limited and is likely to change over time.

Over the next eight years the government, through its NBN Co, intends to build a fibre-to-the-premises (FTTP) broadband network, and sell wholesale access to that network to other companies, who will then use that access to provide their services to businesses and consumers. The fibre to be laid can handle much higher speeds than existing cables, and even though the equipment at either end might need to be upgraded, NBN Co argues that this is still cheaper than laying new cable. The NBN will replace the current Telstra copper network, which is to be dismantled. NBN Co and Telstra have negotiated a deal to migrate Telstra’s operations over to the NBN, while giving NBN Co access to their pits and ducts for the purposes of laying this new cable.

Internet access and telephony are the most obvious examples of services that will be delivered using the NBN, but it could also be used to provide other services, such as television and on-demand video.

Under the Government’s plan, 93% of Australian premises—households and businesses—would be connected to the NBN by optical fibre cable, with most of those premises connected by a gigabit passive optical network (GPON). In a GPON, a single fibre optic cable running from an ‘exchange’ splits into up to 128 fibre optic cables, which are then connected to individual premises.

A GPON provides up to 2,500 Mbps of symmetric bandwidth (that is, 2,500 Mbps downstream and 2,500 Mbps upstream) on each fibre optic cable running from the exchange. However, that bandwidth is then shared between all of the users connected to the same splitter. In practice, most GPON networks split the bandwidth between fewer than 32 users and businesses may be eligible to get direct fibre, meaning that bandwidth would not need to be split at all.

Taking all of this into account, NBN Co is saying it will deliver 100 Mbps symmetric bandwidth to the 93% of Australian premises connected using fibre (and at least 12 Mbps asymmetric bandwidth to the remaining 7% of Australian premises using a mix of other technologies).

Other existing broadband technologies, such as DSL, cable, or wireless cannot deliver the high symmetric speed the FTTP network proposed by NBN Co promises.

Industry watchers took a lot of notice when Telstra announced last year that it would roll out DOCSIS 3.0 on its HFC cable network, enabling speeds of up to 100 Mbps—some pundits went so far as to claim that it meant that the NBN would definitely not be built as a result.

Naturally, people compared Telstra’s 100 Mbps claim to NBN Co’s 100 Mbps claim, and concluded that they were the same. But they’re not. On Telstra’s HFC cable network, the 100 Mbps is split between all the users (often hundreds) who share a single node on its network. As explained above, the NBN’s GPON also shares bandwidth in a similar way—but it shares a much larger pool of bandwidth (2,500 Mbps) between fewer users (fewer than 32, typically). Furthermore, should the NBN’s GPON not provide enough bandwidth between 32 users, it can be upgraded to 10,000 Mbps, 40,0000 Mbps, or higher in areas where this is needed.

Similar analysis is applicable to wireless broadband technologies, like Telstra’s Next G network. Earlier this year, Telstra became the first telecommunications company to provide dual carrier HSPA+ services, delivering downstream speeds of up to 42 Mbps. And further upgrades to 84 Mbps are planned beyond that.

But these wireless speeds are theoretical maximums only: they’re attainable only under perfect conditions and, even then, that bandwidth is shared between several users. And unlike the NBN, the technology is asymmetric, so upstream speeds are much slower than downstream speeds-typically less than 1 Mbps.

DSL is the most common broadband technology in use today. ADSL2+ delivers downstream speeds of up to 24 Mbps, but, while the bandwidth isn’t shared between multiple users as with the technologies mentioned above, it is significantly affected by the distance and quality of copper wire connecting the user to the exchange.

Typical downstream speeds are closer to 10–15 Mbps. And, again, ADSL2+ is asymmetric, with upstream speeds typically closer to 1 Mbps. High-speed symmetric DSL services (such as SHDSL) exist, including some services that use more than one copper pair to get higher speeds, but these services are considerably more expensive than ADSL2+ and they’re available only in limited areas.

There’s also satellite broadband, which is expensive, relatively slow, and suffers from high latency as the signal needs to travel an extra 35,786 km or so to geostationary orbit and back. It’s only suitable for remote regions, where other technology cannot be deployed cost effectively.

Hopefully from the above it is apparent that the NBN offers a significant speed advantage—in particular on the upstream side—relative to other available technologies. But the question remains whether that extra speed is worth the extra cost. What does that extra speed actually deliver?

Recently, Shadow Communications Minister, Malcolm Turnbull, remarked that ‘for most, if not all applications, much lower speeds are perfectly [fine]. If you could deliver nationwide 12 Mbps at relatively modest cost compared to the NBN, what is the additional utility [or] value of going from 12 to 100  [Mbps]?’

That view has some support, particularly with the news that only about one in ten households recently offered connections to the NBN in Tasmania took up the offer. And of those households that did choose to connect to the NBN, only a small minority opted to pay for the maximum 100 Mbps speed. It seems that the demand for 100 Mbps connections simply isn’t there.

But all of this misses the point: the NBN isn’t about the bandwidth that today’s residential users need to use today’s consumer services. With an estimated build timeline stretching into eight years, the question isn’t what bandwidth Australians want today, but what bandwidth Australians will need to have to be globally competitive eight years from now.

There are certainly many consumer-level services that could benefit from higher bandwidth. Anything that involves streaming high-definition video on demand needs a lot of downstream bandwidth. The amount of bandwidth needed depends on the resolution and the compression quality: Blu-ray quality 1080p video, for example, has a bit rate between 15 and 40 Mbps, though lower quality video needs substantially less bandwidth.

The Web 2.0 revolution—YouTube and the like—has meant that so-called consumers no longer just consume. They want to create and share content, including video content, with family, friends, and the world. And at current upstream speeds, sharing anything but short or low-definition video is all but impossible.

But it’s really businesses that stand to gain the most from the NBN (except those whose monopolies might be eroded by it). It’s a myth that businesses already have access to ubiquitous, high-speed broadband. While some businesses, particularly those in the CBDs of our capital cities, can get high-speed, symmetric (but expensive) internet access, most business can only get consumer-grade broadband (particularly small businesses in the suburbs).

Anyone who’s ever needed to access a large PDF file from a small business branch office (particularly if they didn’t know which large PDF they were looking for) will tell you how frustrating it is. Because such offices are often connected with asymmetric broadband, like ASDL2+, the speed at which users can access files from other offices is often limited to around 1 Mbps.

The other emerging trend, particularly in business, is towards more cloud computing. It’s a buzzword but the idea is simple. If you’re running a doctor’s office, an accounting practice, a law firm, or just about any business, you’re not in the business of managing servers. Yet many such businesses have to maintain an assortment of servers in order to operate-mail servers, database servers, file servers, and so on.

In many cases, it is more efficient for an IT services company to run these services for a number of businesses in a centralised location with specialised staff, rather than each of these businesses maintaining their own servers.

However, those businesses need to have the high-speed symmetric Internet access to use such hosted services. Most local area networks-the networks that connect computers and servers within a single location, like an office—run at 100 Mbps or more. A 1 Mbps upstream connection, such as that available to many small businesses today, is no substitute.

It is clear enough that the NBN will deliver benefits to businesses and to consumers, the question remains whether it’s worth the cost.

The Government’s Implementation Study estimates that $26 billion of Commonwealth funding will be required to build the NBN. The rest of the funds needed to build the $43 billion network will come from the private debt markets and, once part of the NBN is operational, from NBN Co’s revenues.

In round figures, the $26 billion Commonwealth outlay works out to about $10 per Australian (or $30 per household) per month over the eight year construction period. It’s important, though, that NBN Co will generate revenues from the network, and that the Government presently expects to sell off NBN Co after 15 years with a 6–7% return on this $26 billion public investment.

It’s a question of policy. The NBN will provide speeds that are significantly faster than those that are deliverable with existing technologies. And that speed advantage will enable new services, both for consumers and for businesses. But I’ll leave it up to you to decide whether it’s worth the cost.

Telstra recently announced that it will upgrade its BigPond cable service in Melbourne to 100 Mbps by Christmas 2009. Big deal.

Somehow, this has excited some people. Alan Kohler, for example, writes, ‘Telstra’s decision to upgrade its cable definitely now means that the National Broadband Network won’t get built’.

Two problems immediately come to mind. First, you’ll never see those speeds. Even if you’re in an area that can get the upgraded cable service, the 100 Mbps is shared between a number of households. You can get the full 100 Mbps only if none of those households is using its cable broadband.

Second, if you’re ever lucky enough to see those speeds, you won’t see them for long. BigPond’s $139.95 cable plan includes only 60 GB of usage (uploads and downloads). At 100 Mbps, that lasts 80 minutes. Excess usage is charged at $1.88 per second.

Incidentally, Next G has the same problems, except that it’s slower and even more expensive. At its peak speed of 21 Mbps, the 10 GB you get for $129.95 lasts only 63 minutes. Excess is charged at $0.66 per second. Keep that in mind next time Telstra brags that its Next G network provides 21 Mbps to 99% of the Australian population.

Fast speeds seem to be more important to broadband users than large quotas. Perhaps that accounts for how little has been said about increasing download quotas in the context of the National Broadband Network debate.

But it should be obvious that increasing speeds without increasing quotas won’t enable the bandwidth-intensive applications that the government hopes to enable.