It is that time of the year where writers look back on the year just passed and bring out their predictions for the future. Today, a look back at a popular subject this year: Network Architecture.
Image: Ricardo Gomez Angel on Unsplash
As I sift through my social media engagements this year (views, likes, comments), network architecture was one of the significant buckets. Why would this be so?
You the reader would know better than me, so feel free to comment, but I would speculate the following:
The position of IP architect is viewed as being further along a career path than a network engineer
Architecture is a form of strategy, network professionals are problem solvers. Anyone did not watch the Queen’s Gambit yet?
It is perceived as requiring more intellectual engagement than just knowing how to configure something
It holds out the promise that it will make everything alright, it will put the dams and dikes in place so the water never comes rushing in
As I wrote previously, architecture is akin to strategy: “The fundamental responsibility of network architecture is to define success in networking terms”. The fundamental role of any strategy is to frame the conversation, what are the most important things to consider, what are the options, what is the definition of success, and how resources could be allocated to achieve that definition of success. This sounds to me like network architecture.
Framing the conversation: what are the business goals, what does this mean in networking terms, how old is the equipment in the network, what can it really be expected to deal with, what are the unsolvable problems (speed of light,…), where should the focus of network design be to account for equipment capacity/capabilities, operations capacity/capabilities, budget constraints, and desired outcomes.
According to Wikipedia, a 1 century AD treatise on architecture proposes that a building should have three qualities:
Durability – a building should stand up robustly and remain in good condition
Utility – it should be suitable for the purposes for which it is used
Beauty – it should be aesthetically pleasing
I am writing a little bit of software at the moment, so I probably, in that context, tend to see durability and beauty as being closely aligned: if I can maintain a simple and easy approach to multiple tasks, the maintenance will be easier, and I am more likely to see it as a thing of beauty. If the way I am coding something is getting too convoluted, I drop the feature, simplify everything else, and wait for a solution to present itself (ok, I have a luxury that some may not have). That said, beauty in software can be a powerful algorithm that cuts many lines of code and executes quickly. Beauty can also be stylistic, as it can with anything.
Durability and utility are for sure important qualities of network architecture. I suspect most networking professionals would advise against pursuing beauty at the cost of utility and durability but see the previous paragraph.
When it comes to getting s*&t done, there of course is nothing more valuable than the engineer who knows exactly how a piece of equipment or software will behave in response to a given configuration or set of conditions. That is a given. There is no guarantee in networking that any two routers will behave the same way in the presence of essentially the same configuration.
In economics, there is great concern about how the big levers of macroeconomics (fiscal policy, monetary policy, stimulus, public vs private enterprise, …) will impact what happens in microeconomics, how much and how often will someone pay for a Lyft ride, etc. People who have studied economics proudly proclaim whether they like macro or microeconomics.
Network architecture is that way of thinking that says if we put these information firewalls in place, the control plane won’t meltdown when something goes horribly wrong, mostly (1); that if we put this server ten miles from the consumer, the latency will be much lower than if the server is in a centralized location; that a given approach to networking can deliver the business requirements, services, and quality of experience.
Architecture is the macro conversation that frames the micro conversation. It is the first line of defense as well as the first line of offense. It looks at the challenge holistically, breaks it down, identifies threats and opportunities, it points in the direction of how resources should be allocated. It is to a network, what business strategy is to a business. It is the thing on top of which everything else is built. It is the foundation. That is as a good a reason as any, to care about it, and one reason I suspect many do. Do you care, and why? Post a comment.
Notes:
(1) I say mostly the control plane won’t go down because it is difficult to foresee every entanglement. A few decades ago, a line card failure in a voice switch bought down a good chunk of the eastern seaboard in the United States. How that failure would be propagated to other switches over and over again as they cycled up and down, was presumably not foreseen.