Run a simple Google search for "pace of innovation" and take your pick among the charts, graphs and point-of-view papers that show. It's a train that's never slowing down to let anyone on, and end-user demand will never truly be satiated.
In chasing the next big thing, application developers, service providers and other organizations are rolling out new services requiring a back-end that's getting more and more complex by the day. For instance, microservices that segment parts of a service into smaller clusters in the cloud have proven to increase development velocity and agility, but managing that mess becomes a whole new ballgame.
Of course, end users never see any of that (and they shouldn't) but to thrive in an era of rapid disruption -- even among smaller companies that don't have the resources of a Facebook, Google or Amazon -- automating is an absolute must.
With the near future of self-driving cars that require ultra-low latency, an industrial Internet that predicts machine failure and automatically assigns fixes, or even just advances on the mobile network like 5G connectivity, network automation is going to start cropping up in whole new ways.
As new IT challenges emerge, here's a rundown of the expected new automation innovations -- and challenges -- that network operators will need to consider over the next five years.
Reallocation of human capital
With automation comes perceived threats to job security. But network automation isn't going to replace anyone; humans are still a necessary part of the equation. Rather, it's going to augment them while freeing up time to focus on more critical tasks.
Few network operators -- even those who have learned some programming -- are knowledgeable about the ecosystem of DevOps tooling, and few consider applying those same tools to networking: DevNetOps. Learning Python and Ansible is just scratching the surface. There's an abundance of DevOps tools for CI/CD, site reliability engineering (SRE) and at-scale cloud-native ops that are begging to be learned and applied in networking.
The way most networks are constructed, humans still need to provide the intent, but that is rapidly changing as intent-based networking catches on. The right automation tooling integrated with an existing monitoring system can enhance the quality, speed and accuracy of incident management, for networking or in general. In telling a network what they want -- not how -- IT staff will need to master the concept of intent and telling a system the outcome they desire.
Re-training staff to improve their automation and provisioning prowess is a must in the automated future, shifting from manual to automatic; from configured to coded; from change-controlled to continuous; and from tracing and rummaging through minutiae to drilling down through layers of abstraction.
Automation as an enabler
As data crunching moves from devices to the distributed cloud at the edge, the way we think about monitoring application performance must also change to ensure real-time processing for seamless low-latency applications. Intelligent software using machine learning and artificial intelligence will respond to issues and fix them in real time -- not minutes or hours -- to enable network operators to better manage applications and services that rely on their networks.
Networking administrators have been wrestling with proprietary systems and infrastructure for years, inhibited by manual troubleshooting. Time is of the essence, and seconds count when delivering services to end customers.
Edge processing and automated service assurance will eliminate any latency and ensure speed for billions of users.
The move to automation is going to mean not only up-front capital investments, but also about twice as much as that in terms of integration on the back-end to ensure the systems all work together to eliminate manual tasks. When network operators begin leveraging NFV to bring in network functions on an as-needed basis and programming is polyglot, it can bring more complexity than it's seemingly worth.
To help alleviate this, use APIs and standards-based protocols that reduce integration and interoperability issues with third-party applications and back-office systems.
Once integration is complete, automation will bring consistency -- and simplicity -- to the code such that the endeavor makes operational and business sense.
Machine learning and AI
Virtualization brings the ability to flexibly cater to many different end-user needs, but human intervention to maintain all SLAs is nearly impossible at scale. Only automation can enable zero unplanned downtime across the spectrum.
Using telemetry to provide rich experiences to customers while still maintaining operational excellence, machine learning and AI will help in detecting anomalies using predictive analytics to correlate and combine data based on things like geography and traffic patterns.
We aren't there yet but are starting to see promise in the industrial space where predictive maintenance that avoids unplanned downtime with big machinery is catching on.
A majority of companies in the market are in the very early stages right now of replacing manual tasks and operations with automated systems. Automating the whole network is going to take time and probably happen in bits and pieces. According to McKinsey, it will likely take until 2055 to automate just half of today's workloads. So for a while, humans are going to absolutely continue being a critical part of the equation.
But automating networks will pave the way for the future of services -- some that can self-assemble in the background to change the way we live. Networks that require manual inputs simply won't work in that kind of future.
We need to start now, and successful change requires concerted and sustained effort, not a patchwork of disconnected responses after the fact. Ultimately, organizations must make sure that their network becomes self-learning and automated to ensure that they thrive in the digital age.
— Sumeet Singh, vice president of engineering at Juniper Networks and founder of AppFormix