Thinking about Computer Networks with Ed Birrane

Artech House author Ed Birrane, whose title, Delay-Tolerant Applications: The Role of Store-and-Forward Networks in an Increasingly Real-Time World, gave us some insight with regard to the future of computer networks:

We need to think differently about computer networks.

It is not easy to have a casual discussion about computer networking minutia. If you doubt this, I suggest timing how long you can maintain a discussion on acknowledgement strategies at your next family birthday gathering. If you best 5 minutes, let me know. Harder than “geeking out” on as-is networking is to discuss why what we are currently doing with computer networks is not enough. Try explaining to the lay-person that “The Internet” – the most impactful communications mechanism in the history of our species – isn’t going to keep working unless we think about networking differently. That is not just a boring conversation, it is also depressing and perhaps controversial.

Beyond the birthday party experience, I have repeated this conversation multiple times in government, industry, and academic forums. And the conversations are no less uncomfortable: what do you mean our business model/framework/strategies will not scale? The thought that your last multi-million-dollar investment only buys you a few years of comfort can be far more depressing than delayed birthday cake.

But we must learn to develop networking applications differently than the ones working so well today. How can we talk about this in a more understandable way without delving into point solutions? I have had some success with the time-honored tradition of using incomplete analogies to describe complex technologies. Let’s try one here.

Consider a typical sportscar advertisement: a driver speeding down an empty highway… aerial shots of windy roads and breathtaking vistas… interior shots of said driver focused on the unencumbered joy of getting from here to there.

Not a representation of New York City at rush hour.

That advertisement is an escapist fantasy. It does not represent the actual bumper-to-bumper-across-five-lanes-of-traffic rush hour experience[1] enjoyed by most users most of the time. Yet this same fantasy is analogous to how we reason about packets – speeding through an otherwise wide-open network. And if there is some congestion we can “ride it out” with timeouts and retransmits until the network comes to its senses again.

But what happens when the network never comes to its senses? What if the network is always congested, or contested, or covering astronomical distances? Wrapping up my imperfect analogy – if you live in downtown New York City, maybe you think about transport differently. Maybe you don’t buy that Ferrari. Maybe you buy a bicycle.

As we instrument our world, our networks will behave differently.  Billions of users want 4k live video streams and wireless media delivery. Billions of machine-to-machine connections and IoT devices require steady-state, round-the-clock bandwidth. As we instrument increasingly remote parts of our nations, the world, and the solar system our networking infrastructure will see demands unlike ever before. New physical layers, more cell towers, and data compression all buy us time keeping our networks running as-is. But they don’t solve this problem at scale.

I wrote Designing Delay-Tolerant Applications to stimulate thinking about networking application development. A delay-tolerant application does not require a high-availability, low-latency, end-to-end network to operate. So much of our data does not need to be real-time – if alternate networking paradigms existed for non-real-time data, we could dramatically reduce burdens on our networks by rethinking our algorithms and workflows.

This book discusses the motivation for delay-tolerant, and provides some design patterns for thinking about application tolerance in a variety of networking scenarios (but with specific attention to space-based use cases). The ultimate value of this work, I hope, is that is makes you think about networks that never “hit stride”, that never have enough bandwidth, and what you might be able to do about that.

[1] I long for the ad agency that extols a top speed of 65mph, a massaging seat, an audio teleconference suite, and a high capacity fuel tank.

Click here for more information or to order.

Leave a Reply

Your email address will not be published.