I came across this today:
I’ve been working on a software project for the past 18 months now and am in the final stages. This involves setting up our own servers and while some might think that’s technically challenging — the hardest part has been figuring out how to hook it up to the internet. Internet in Australia sucks.
For reasons I don’t know, I tried commenting this in small segments but can’t seem to so here it is instead. Here’s the original post for those that can access it — https://www.facebook.com/groups/sydneystartups/?multi_permalinks=2019137678127502&comment_id=2019550068086263¬if_id=1529112936582794¬if_t=feedback_reaction_generic
While federal governments have been responsible for overseeing large infrastructure/civil projects before, train-lines, Harbour Bridge/Opera House and even airports, these are all a type of project that is quite technically different to that of installing a nationwide high-speed internet network.
There’s been some really good responses so far but not many have touched on why it’s such a technical challenge to achieve consistent high-speed service. That said, this isn’t an excuse and I do think Australians are being effectively held hostage by internet-service providers, namely Telstra and Optus.
Firstly, let’s remind ourselves how we access the internet. For those unaware, you could visualise internet as just like the folders on your desktop. Imagine each folder is a website and so with this analogy we can start to see how the internet is just a collection of folders that are accessed on different computers — web servers. So when I went to open Facebook, I sent a request to Facebook that sent back the content from their servers and then that was rendered on my device.
We send this data through packets of information, for reference — there were ~ 420 requests and 382 KB transferred when I refreshed the “Sydney Startups” Facebook page.
To connect my device to the server, we need a link. Overlooking the underwater cables that connect Australia to North America/Asia — https://www.submarinecablemap.com/#/ — we have a series of connections between different devices/amplifiers that moderate and change the signal so it can be processed by our device. The primary connection can be hybrid fibre coaxial cable, digital subscriber line (DSL), integrated services digital network (ISDL) and satellite.
Most of these rely on copper wiring and this is the root to a number of issues. For reference, copper wires are naturally the strongest type of conductors — highest electrical conductivity rating of 16.78 nano-Ohms at ~20-degrees Celsius, and atomically wise, copper’s atom is only half filled with electrons so, because an electric current is the motion of charges, copper can carry a lot of electrons (a current is a measure of electrons over the rate of change of time (dQ/dT)).
We’ve just touched on the most important part here — the need for electrons to be in motion for an electric current to be established. This is the Achilles heel of modern internet infrastructure.
Now consider the state of Australia’s internet infrastructure today. When it was realised in the 80s/90s that the internet was the next “thing”, telecommunication companies knew that they had to switch fast and thankfully internet can be transmitted via the copper wires that had already been laid for telephony networks. Telecom companies have always been the bread and butter of modern societies — take the American Bell Telephone Company, they were broken up in the 80s (leading to AT&T )because they owned pretty much all the telephone exchanges in America. A key reason to their success was their “first-in” position which can be accredited to Alexander Bell — he invented the telephone so there was literally no competition. Installing telephone networks/exchanges across the States is an expensive operation due to the physical costs/labour so the faster you can connect 1, the connection is now established, it’s not like providers can add new lines or nodes and 2, switching is an inherently intrusive/taxing procedure so there’s less chance of losing customers.
AT&T’s dominance of America today is pretty akin to Telstra’s. Telstra owns 60% of the copper wires within Australia’s telephony network with the second primary owner being Optus. This means that if you want internet access in Australia it most likely has to come from Telstra or Optus because they own the copper wires.
On a personal note, this duopoly has meant a complete starvation of flexibility for consumers. Take for instance my own back-end servers — these require a high-speed synchronous connection so it can both receive and send back data for clients. For testing we planned to use my own network at home that’s hooked up to an Optus connection — we usually have decent speeds at ~145 Mbits/s. Unfortunately Optus blocks port-forwarding meaning there was no way we could access WANs which makes a lot of testing impossible. The only option was to consider a business plan — another contract, more $$$ and all for a restricted, fixed bandwidth.
This is just one example of pretty wacky reality — Australia is one of the few countries that have internet access regulated by the amount of download, within OCED countries we have the highest $/MB after exceeding cap (0.103 USD/MB, the second is Ireland at 0.018 USD/MB) and worst of all — our internet pricing is 20–21x more expensive than Europe.
I’ve yet to hear a solid reason why 1, internet is regulated by download-amount and 2, why we pay so much.
This realisation were some of the motivations for the NBN with the initial goal being 1000 Mbit/s connection and using optical fibre for fibre-to-the-node or close derivates of it. As we’ve mentioned, current internet performance is partly a function of the rate of electrons and so it isn’t a surprise to learn that the physical properties of copper/environment surrounding copper impose limitations on performance. This means that for local-area networks, copper lines can support ~3k phone calls at once as opposed to the 31k that a fibre-optic can.
The reason for this is simple — fibre-optics use light. Light travels at 300,000 km/s and as we’re dealing with photons/elementary particles — we’re not limited to the same sort of constrictions that come with copper wires. We could really delve into this but that’d be venturing into quantum mechanics which is unnecessary.
Fibre is also impervious to electromagnetic interference — because it transmits information via light. This leads to significant benefits: a copper wire can usually transfer a signal for 100m at 10Mbits/s whereas fibre-optic can do the same but 40km and at 69 000 Mbits/s.
However we now have the problem of this awesome technology but replacing copper wires is something of a ginormous task. 80% of the cost for hire comes from installation/labour. The period of time it takes to break even for digging into concrete/installing wires/then resurfacing a road is literally years — and that’s for every few metres of fibre.
Before we start grilling NBN, let’s check out another attempt that by all means should’ve been able to do it. Google.
Google Fiber started in early 2010s as an experiment and once it had demonstrated success/viability as an entire business, it began rolling out across cities in the States. These were the options:
1. $50 USD p/mth => 100 Mbit/s (synchronous), no storage
2. $70 USD p/mth => 1 Gbit/s (synchronous), 1 TB drive
3. $160 USD p/mth => 1 Gibt/s (synchronous), 1 TB drive
There was no installation/construction fee. However, as soon as it started cracks began to show. One employee went on to say this: “Everyone who has done fibre to the home has given up because it costs way too much money and takes way too much time”.
It’s a gruelling physical process — when a new provider wants to attach their wires to a house, each provider with wires already on the local pole has to send a technician to move its wires to make way for the new ones. No surprise — AT&T and other providers took on average 3 days to just respond to a request from Google. Regulation meant that Google fibre was able to only upgrade 33/80 000 poles in Nashville.
For those curious, this is the document Google sent to cities to prepare for installation. https://fiber.storage.googleapis.com/legal/googlefibercitychecklist2-24-14.pdf It’s an explicit insight into the sort of concerns these companies have when trying to install fibre.
As a consequence of regulation and that it cost on average $1 billion USD to install fibre per city, Google Fiber is pretty much dead (pretty sure it’s still available to the houses that connected but the company isn’t expanding their service).
The primary cause of this is the cost for using fibre-optics. Anything underground based is inherently costs from both installation to operations and has the added downside of severely restricting optimisations/performance improvements. The internet performances of today is a derivative of the telephony performances that were established in the mid 1900s. Ripping out the existing copper network and replacing it is ludicrously expensive and also has low profit margins (usage/square kilometre).
So what’s the alternative?
Presently not much in Australia given the only way to access the internet is via the network provided by Telstra/Optus.
That means alternatives need to look into ways of bypassing these networks which lead to a few options: 1. Free optical space communications — basically using optical-based communications via air mediums (somewhat in R&D phase at the moment, nothing at a commercial scale soon), 2. Greater wireless communications — wifi operates at 2.4 GHz and 5 GHz frequencies with each of these having pros/cons. The primary issue is the strength of the signal — 5GHz signals can’t penetrate walls as well as 2.4 GHz which already has its own issues. 3, My personal favourite and pet-hobby-idea is LAN drone-based internet via mesh networking. SpaceX is kinda thinking this with their internet-via-satellite plan but its feasibility/long-term viability as providing global internet should be seen as more as an experiment for now rather than “the” solution.
Lastly, another alternative is to bypass government projects, I wrote about why here — https://medium.com/@bryanjordan/in-1898-samuel-langley-received-70-000-4424e37adb8c