The next ARM is a networking company

The smartphone revolution is only possible because we broke the trade-off of low power consumption and high performance. This was achieved through the application of superior science and elegant engineering. A similar jump awaits the networking industry, but only if we can nurture the new mathematics and architectures from obscurity to the mainstream.

Back in the 1980s, many of us were caught up in the excitement of the home computer revolution. I still have the tactile feel of the BBC Micro power switch and keyboard etched into my muscle memory. The devotion to simplicity and elegance led people like Sophie Wilson et al to the ARM architecture.

I had an ARM3 in my undergraduate bedroom in my Acorn Archimedes A440, with its 4Mb of RAM and 40Mb hard drive. I used to write ARM assembler as an antidote to functional programming on my official syllabus. I had worked every school and university holiday doing VDU data entry and washing dishes to save up the £2500 pounds (at 1990 prices!) to buy all the kit.

Three decades on, and we now live in a networked world. Home computing has now become distributed computing, delivered in many forms and formats. We might think that data networking is a “solved” problem, but I believe we are about to experience a hard-to-imagine leap in performance, capability and cost due to some recent breakthroughs.

These offer the “RISC architecture of networking” that nobody was aware was even possible. This impending leap has two sources of inspiration. They are unified by a single vision of networking as being interprocess communications (and nothing else).

With computing, people like Turing and Church put the theoretical foundations of computation in place before we built digital computers. With networking, there has hitherto been no fundamental theory of “information translocation”. That gap as been fixed with a new branch of mathematics (see qualityattenuation.science).

So what? The result is that we can now precision engineer application performance whist running networks at >100% load (with graceful degradation). Or in other words, we can drive networks to their theoretical performance limit. (Most access networks today are <1% loaded and nobody can assure application performance at any reasonable cost).

The Internet is based on the TCP/IP architecture that essentially got everything wrong apart from one thing, which was to abstract connectivity. The RINA architecture approaches the problem from a completely different perspective, and outstrips TCP/IP in every possible way.

So what? The result is a huge reduction in complexity and power use, and equally large increase in performance and security. It is simply impossible by design for many of the current Internet’s failures to happen.

The former is like the leap from classical to quantum physics (bandwidth to stochastics); the latter is like the leap from breakbulk to container shipping. There are two simultaneous paradigm changes, which makes this an intellectual challenge to master. This matters, as the constraint on technical progress is now the network, not the processor.

The combination is a jump that is hard to comprehend as being achievable, rather like describing a supersonic aircraft to someone who has only seen a steam engine. Yet it demonstrably is real, as there are working prototypes.

On top of this, there is an impending ‘lean quality’ revolution in telecoms that matches those already undertaken in primary and manufacturing industries, as well as service industries that operate at below the speed of light.

All of these ideas are rapidly maturing to being product-ready after many R&D and early commercial uses. Yet what is critically lacking are patrons of the pure science and mathematics. I and my associates have spent years creating public science goods at private expense. Spreading the gospel of redemption via rigour is slow and costly.

What is now needed are people to help me and my colleagues to complete Turing’s work. This will help society to get the full benefits by being informed of what is possible, and how to achieve it.

I cannot imagine a greater legacy than leaving our grandchildren a radically better Internet that is fit-for-purpose for any application you can conceive of – just like ARM is for computing. Who knows, maybe some day the BBC might even run a Networking Literacy Project on these breakthroughs?

For the latest fresh thinking on telecommunications, please sign up for the free Geddes newsletter.