Future Tech 2024: An Interview with Dr Paul Brooks (Layer 10 Advisory)
In this Future Tech interview, we’re speaking with Dr Paul Brooks from Layer 10 Advisory. Often described as a “Serial CTO,” Paul’s expertise in telecommunications network design, planning and operation stems from extensive hands-on experience in broadband access and large-scale data networking with leading equipment suppliers, carriers, service providers and regulators. His latest project is with Trident Subsea Cable, a $400m subsea and terrestrial optical fibre network linking Perth with the Pilbara/North West Shelf and internationally to Jakarta and Singapore.
Shara Evans (SE): Today, it’s my pleasure to be speaking with my good friend, Paul Brooks, of Layer10 Advisory. Paul is a noted technologist and the architect of many of Australia’s broadband and networking standards.
Paul Brooks (PB): Thank you, Shara.
SE: I’d like to start our discussion by talking about broadband access and technologies, starting with developments in Australia.
Consumer Broadband in 2024
SE: If I look at the telco world in 10 years, what do you think a typical consumer will be using?
PB: Well, much what they’re using now but maybe a different mix of technologies. Consumers are now using many different types of technology depending on where they are or what they’re doing — sometimes many different forms of technology at the same time. For instance, when they’re sitting on the couch and watching the telly with their tablet in their lap, then they’re using a form of WiFi but it’s generally coming off a fixed broadband connection, so there’s a mixture of two that people tend to look at.
Really, it will still be dependent very much on where they are, whether they’re in the city or in the country or out in the bush, in Australia at least — with maybe a different proportion of mix of high-speed DSL services as the NBN rolls out. Some of the country will have fibre connections; some will have copper connections on VDSL. Maybe a little bit of fibre to the distribution point with G.fast that will be commercialised and maybe a little bit more widely deployed by then.
So it will be a mix in 2024. It’s a mix now. It was a mix 10 years ago.
SE: With G.fast in particular, how fast do you think it’s going to go?
PB: It could go up to several hundred megabits. Interestingly, for a residential broadband technology, that will be almost symmetric and provide very good upstream capacities, as well as downstream. If the distribution point where the G.fast DSLAM is literally at the bottom of your driveway, then over a few tens of metres of your telephone cable — and with no other cables around it providing too much interference, then several hundred megabits per second should be quite doable.
SE: In what kind of time frame do you think that would be doable? Is that something that is going to be commercially available very soon from the vendor perspective, not necessarily from a deployment in Australia perspective?
PB: Well, certainly within the next 10 years. The standards are due to be ratified in the next year or two. The chipmakers have silicon implementations in trials at the moment. And, there are some vendors that are rolling out here using pre-standards equipment.
If we look at the typical lifestyle for a communications standard to go from the theory to ratification, through the manufacturing process, and end up in equipment that you can buy off-the-shelf of the store, that’s typically a three- to four-year program. So in four to five years, I would expect that we will see equipment off-the-shelf that will be relatively expensive, but provide these high speeds. What it will wait on though is the network operators to actually put the equipment out in the streets, in the basement of apartment buildings or wherever, for those sort of modems to connect to.
SE: I suppose the NBN could play a part in that.
PB: The NBN can certainly play a part of that or may well do. G.fast and FTTdp, as it’s called, was part of the mix of technologies that the Strategic Review recently looked at. Once the standards mature, you can imagine that in the final stages of the NBN rollout, as a bit of a compensation for those that end up coming last in the rollout, the equipment may well be cheap enough and deployable enough in bulk for that to be a viable technology for those areas to get.
SE: So it may actually be better to come in a little bit later in the game rather than earlier in the game…
PB: It could easily be.
SE: …unless of course you’re getting fibre to the premise, in which case you’re probably a very happy customer.
PB: Then you’re already well ahead of the game.
The Australian technology landscape: How will it compare to the rest of the world?
SE: Now from a technology mix, do you expect this to be different in Australia as opposed to other parts of the world?
PB: Yes, I do. Typically, you really divide these cities into the highly the concentrated urban areas, similar to many of the Asian cities in Japan and Korea, or Hong Kong, where you have a very large number of apartment buildings, as compared to cities like Australian capital cities and regional areas, Canadian cities, many US cities, which are typically more spread out, more suburban, less high-rise, with more single-dwelling houses.
The technologies you typically use for those sorts of things are quite different. You would expect to see countries with cities that are constructed similar to Australian cities having a similar mix. Whereas in the high-density apartment-block-type cities — those apartment blocks will typically use a completely different technology, maybe fibre or maybe Ethernet higher-spec copper cables up the risers to provide the gigabit-type services that will be required in a decade’s time.
SE: Thinking about Australia, we’re pretty remote from the rest of the world. What kind of role will our international connectivity play in this whole area? Perhaps you can touch on data locality and what caching networks mean.
PB: Yes, this is a really important one. A lot of people don’t realise that however fast your access link to the Internet is, if you are trying to download or upload information to a server that’s a long way away, the limitations of the Internet protocols will slow that down, independently of your actual link speed. So while it might be fine to have 100 megabits or 200 megabits on your link from your house to the Internet, the reality is that unless you do something special with the protocols, if you’re trying to download something from the US, it’s only going to run at somewhere around 10 to 15 megabits per second.
Now if you run 5 or 10 of those, or you have multiple people in your house doing things at the same time, then you will end up using up that bandwidth. But a single person, because of the speed of light delay, can really only download something at a certain network speed. Now the closer that server is to you, the faster that transfer will occur. So one of the things we see in Australia already are content-distribution networks locating servers within the borders of Australia and serving up the content from there, so that advertisers and customers can get a good experience — comparable to what is seen in other parts of the world.
Now this is also true between, say, the US and Europe; or in Asia where the distances are typically much smaller. Sometimes you get some caching locally in one country — say, Singapore or Hong Kong— and not necessarily need a great deal of local content caching in a close-by city. But trying to get around that speed of light delay is something that is a basic law of physics, and it doesn’t matter how much fibre and how fast you run the fibre, you still can’t get around time delays.
SE: No, we haven’t broken that law of physics yet!
The other thing that comes to mind in this equation is the speed of the connection used by the content hosting server, because even if you’ve got super-fast fibre at the access network and in the international backbone, if the server itself only has limited bandwidth, and a lot of users trying to access it, guess what, you get slow speeds.
PB: Well, sure, but we’ve got to the stage now where the hardware is so cheap that putting in multiple servers, tens, hundreds, thousands of servers, to satisfy the load, it’s not all being fed from one machine, even if they’re all located in the same data centre.
The advances in local network technology with 10-gigabit, 100-gigabit Ethernet, and 200, 400-gigabit is coming through the labs — terabit Ethernet connections to a single server are certainly on the drawing board for, I’d say, within five or six years.
So, that side of things is less of an issue. It’s still mainly the distance and the speed of light time and the performance of the networking software within a single person’s PC — particularly if they’re accessing it over a WiFi connection using a handheld device. That’s where the bottleneck will be.
It actually won’t be in the international cables, interestingly enough, because one of the things with content distribution and caching is that with the content distribution caching in place, a specific bit of content typically only has to be carried into the country once or a small number of times. Whereas if that caching isn’t there, then every single customer drags it over the international links, causing a lot more load on the international cables.
So I’m thinking that the international load on the cables coming in and out of Australia, although it will certainly grow, won’t grow as quickly as people might expect from the growth of access network technology and average speeds.
SE: It’s an interesting observation. I’d also point out that, from a commercial perspective, the cost for international capacity keeps coming down as the cable operators continue to upgrade the speeds of their backbone networks.
PB: That’s right
SE: The cost per bit just keeps on going lower and lower.
PB: The cost per bit keeps going down lower and lower, but the traffic demand keeps going higher and higher. So there’s always a need for new cables, if not for the bandwidth but also to increase the resilience by creating multiple paths. Even if you have the same amount of traffic, being able to spread that out over three cables rather than one gives you a much greater margin of safety when you get an undersea earthquake and a cable or two snaps.
So a lot of the build-out of undersea cable infrastructure at the moment is happening in areas which are prone to undersea earthquakes, and where there’s a need for a carrier to have multiple parallel connections over different geographic paths as a form of protection, as well as increasing their capacity to the rest of the world.
SE: It’s a really interesting trend and certainly we’re seeing a lot of investment, or mooted investment, in submarine cables in and out of Australia.
Business Connectivity Requirements
There’s one other trend that I want to touch on with respect to broadband technologies, and that has to do with the difference between what we’ll see consumers using and what we would expect in small and large business. For instance, the symmetry of connections in a business environment might be different than a consumer environment, although all of these new cloud-based services are really changing the paradigm.
PB: It depends very much on the nature of the business. Again, we’ll get back to geography and where precisely they’re located, because it will be different technologies that are rolled out in business areas, around business parks where technology companies and non-technology companies such as factories, warehouses and things — tend to congregate, than around the residential areas where the focus is on being able to hook up many hundreds, if not thousands, of relatively low-bandwidth services, compared to business park areas where — for a start — the blocks of land, and the distances between buildings and the size of the buildings is so much larger. A distance-limited technology like DSL on copper will only be able to service a relatively small number of actual buildings, plus you get the aggregated demand across the entire building.
Now for a small business — perhaps a fish-and-chips shop, a bookshop, that sort of thing — then a residential-style service may well be good enough. But typically what you’ll see in these sort of businesses is that even if they’re not hosting their own website, they still need reasonably good outbound bandwidth to upload content to their website, and keep content fresh for the website, which might be hosted in the cloud. Or, to access their internal database of stock levels and things like that.
Increasingly we look for real-time instantaneous information for what the actual status is of an item in inventory or in the warehouse, and for the retailer to be able to get that out in real time to the people that are browsing the site or making the query. So outbound bandwidth is very important for those people as well.
SE: Yes, and they’ll be taking advantage of backup storage in different remote locations for disaster recovery or just for collaboration.
PB: Backup is an incredibly important and under-looked application for broadband access.
You see so many small businesses in particular —90-odd percent of the businesses in the country employing 60 or 70 percent of the population — and so often they look for fast broadband for downloads and things like that, and then ignore their internal server customer database backup strategy, which might be going to a USB-connected external hard disk which sits on top of the main server. In the event of a break-in and the server gets stolen or a fire, both the backup unit and the server get stolen.
SE: That’s right.
PB: Now having a routine backup strategy where everything just flows upstream at a broadband link to a completely offsite location and with a fast enough link that should a disaster happen, you can suck it all back down again onto a fresh server and get yourself back up and running in a few hours is absolutely critical to business, and something that today’s broadband technologies really don’t cater for whatsoever.
We’ll be publishing Part 2 of the interview with Dr Paul Brooks on Tuesday 27 May – where we will continue our discussion of telecom networks in 2024, focusing on carrier core networks, encryption and networking in outer space.