The Xceptional Blog

Automation, reliability and trust

Written by Natalie | May 23, 2019 5:24:48 PM

Networks are Complicated and Fragile

Networking is complex. There are many protocols running across the various OSI layers that are highly interconnected. Another way of layering includes the management, control and forwarding planes. In network design, we have various layers including the optical, core, spine and edge. Navigating all of this when trying to gather information or make changes can be rather difficult!

It doesn’t help, of course, that the components used to build the network are fragile. Both software and hardware can be buggy. Knowing what the outcome of a particular change can be difficult, in the face of all this complexity and fragility. And who hasn’t accidentally pulled the wrong cable out of the patch panel?

Organizational Trust and Reliability Engineering

It’s not surprising then, that organizations frequently have a low amount of trust in their network engineers. There is such fear of network issues that often changes are only permitted in a narrow window of time on the weekend. Paradoxically, things are made much worse by this practice, as there are no users around to say they’ve been negatively impacted. And then, of course, the phone will start ringing first thing Monday morning.

With a reliability-first approach to automation, network engineers can help rebuild organizational trust. Instead of using automation to simply perform necessary changes, automation can be used to validate that the changes are working as expected and that they aren’t negatively impacting the network. This new way of thinking about automation – that it’s about reliability and not just speed – will require some creative thinking from the industry as a whole. Figuring out how to validate that the network is working as expected can be tricky. For instance, how do you know that a change in your wireless environment won’t impact the movement of the large number of wireless devices moving around your campus? Problems like this will be difficult to solve, but there is a way. Addressing these challenges will help move the field of network engineering forward.

 

By Derick

Published with permission from forums.juniper.net/t5/Blogs/ct-p/blogs