What's next for Networking?

I am a Network Consulting Engineer by profession. In my job, I will indulge myself in designing, redesigning, migrating and implementing network solutions to help the customer to serve their clients in the most efficient way possible. 

There have been many advances in the networking industry, particularly with the advent of virtualization and Software Defined Networking. What had been a relatively secure and noninteractive technology (almost) suddenly found itself actively interacting with the applications it is ultimately serving. This is the essence of next generation networks; It's not just the data rates that are important, but also the intelligence of how a network caters to different applications, how it secures itself from various applications, is paramount in this information age. Amidst the rapid transitions the networking industry is going through, there are curiously unique concerns and intriguing possibilities of what may lie ahead for us. I would like to explore some of these opportunities and the necessary changes it may bring about in our roles as engineers.

Service driven revenue models

In an open-source software industry no one owns the software, but only custom implementations of it with a long-term support model. Networking industry might one day come up with a similar strategy where the customers can get a particular Networking OS for free but would need to seek support for implementing and maintaining these networks in their demanding business environs. I am implying the presence of 'open' hardware platforms here, although it is not sure if these can scale to the cutting-edge performance offered by leading vendors these days. This service driven revenue models might co-exist with product-service combination most vendors currently offering.

Intelligent Self-Repairing Networks

Imagine a network where each node is intelligent enough to keep track of its health components and the underlying causes for any instabilities. You now have a stimulus and a reaction. If the reaction is negative, you can impose restrictions on the stimulus, if it is positive, you can offer rewards. Now, if a network is made up of nodes like this and if the interactions between the nodes are also kept track of and managed in a similar way, we have an intelligent and self-repairing network, which can respond to any coordinated attacks.

AI-driven Quality of Service

QoS is one of the major components of any network design. After all, a network must be equipped with cutting edge intelligent traffic management technologies (such as MPLS-TE in an SP network). So far, designing QoS requires manual or computer-assisted research of the popular traffic patterns and then equipping the network to deal with such patterns. In many cases, this implementation is static and unchanged even if the traffic patterns change over a period. Imagine an AI-driven system analysing the traffic patterns at a node or a firewall continually and suggesting the optimal configuration for QoS or may be going out and configure itself, thereby eliminating the manual intervention. Of course, unless we trust the underlying algorithms with our life, we may not want the computer to make changes to a billion dollar live network.

Automating Network configurations

Automation is certainly a buzz word in networking where more and more people are talking about automating service configurations, automating resource management, automating network management as a whole. While there are already some versions of automation happening around us in this context, there is a long way to go. The biggest challenge we are facing in automation is the multiplicity of vendors in the market. We need a system which can integrate snugly with current technologies rather than enticing the adoption of a whole new technology.

Self Building Networks

Most businesses do not care about what technology they are using, as long as it is serving their market and cost optimal. And hence in many cases, unless the firm is extremely sensitive to highest possible performance, can make-do with a simple configuration that is secure enough to cater to their needs. Imagine a situation wherein you buy a set of networking nodes, and all you want to do with them is to connect each other linearly and connect the edges to the internet. If you could have a software to represent these nodes pictorially and indicate a few parameters concerning traffic flow, and the software initiates the building of the network and before you know, your network is up and running.

IoT Networking

It is impossible to talk about the future of networking without touching the framework of IoT. IoT is simply about numerous unassuming devices connected to the network, offering us excellent insights into the mechanisms which we often overlook or grossly underestimate. The challenges we currently face in this realm is to come up with a framework of protocols and architectures that can deal with extremely low-powered and memory/CPU sensitive computing nodes. Also, these protocols and architectures need to extremely secure, especially if we are planning on letting IoT seep into our private lives (as home automation, self-driving cars, etc.,)

Offering Networking APIs for Software based enhancements

In my opinion, if there is one component that is highly valued and not provided in the market is the presence of APIs for the networking Operating systems running on vendor-specific or vendor-agnostic hardware. The presence of these APIs, which thus enable critical information retrieval, can enable boundless possibilities to enhance the business intelligence of networks. I am not referring to SNMP or NetConf which can provide or gather information from a node, but a vendor enabled API to provide curated information often in a way the show commands in the command line work.

These are some of the changes that one might see shortly. It may well be happening already around us, albeit in different forms with the same underlying intention. We will see networks which are more agile, more secure and very smart and also easier to maintain.