Image: Pixabay
On one hand, this is a wonderful time to be joining the world of networking, as we have learned so much about how networks work, and we are on the cusp of a new era of augmented and autonomous networking. On the other hand, how often do junior network engineers ask a question, and get the infamous answer “it depends”.
I was reminded of that this week when listening to the Routing Table podcast, What Does A Good Network Design Look Like? - James Bensley, Senior Network Design Engineer @ Sky. James kicks off the discussion with the analogy of a layer 2 aggregation switch feeding into a single router port. The design question is whether it better to do policing (traffic limiting) on the Layer 2 switch or on the router. James goes on to explain that seen through the technical lens of where is the absolute best place to do it, which device does it better, etc. the question is difficult to answer.
On the other hand, if you ask the question, “how detrimental to the service promises the ISP has made, would it be, if there was congestion on the uplink from the switch to the router?”, then the ease of answering the question increases. At the very least, the way the answer is constructed changes, arguably to The Important consideration.
If we step back, a network is a platform, in some ways analogous to a computer or operating system, just with many moving parts that are exposed to the network manager. How a network supports applications of the network, services if you like, is what gives the network purpose and direction. Purpose that is aligned to missions, business models, brand values, customer segmentation, etc.
The intersection of business and technical issues is a fuzzy area. Not surprising that “it depends” is heard so often. No wonder also that the distinction between network architect and network engineer has emerged, so network architects can spend a good chunk of time understanding business issues and translating them into network objectives.
While augmented control planes appear to be picking up momentum, it is likely that augmented planning and design will too. Wouldn’t it be handy if a junior engineer could enter a design question to an operational system, and have that operational system make recommendations based on service requirements, or at least point to areas in the design where the network engineer should focus attention “Hey, that uplink to the router is oversubscribed, which is great from a cost perspective, but has the potential to compromise the strict SLAs that have been committed to customers. Here are some considerations and options”.
Half the battle in getting to good answers, is starting with good questions. There are no questions that are bad in the sense of implying the intelligence level of the person asking them, but there are some questions and considerations that are more on point than others, and have greater potential to remove “it depends” from the answer.
Augmented networking has some specific focuses today, but over time, it is easy to see it trending in the direction of fuller life-cycle considerations. That journey will not happen overnight, but it is a journey worth taking.