In true enterprise IT fashion, we are seeing the beginning of the Great De-centralization. Whether it’s macro trends like distributed cloud, specific use cases like IoT that require an architectural rethinking, or the natural evolution of cloud moving on-prem, it appears that the future might be closer to the edge than previously anticipated.
So, what does life on the edge look like? Here are five steps to living and thriving in a distributed world.
(Image: Photobank gallery/Shutterstock)
Start with Why: For a surprising number of enterprises, the why driving cloud and multi-cloud is still somewhat nebulous.
Cloud is about agility. It’s not surprising then that cloud is simultaneously ushering in the dawn of automation. So important is automation to the successful deployment of cloud that companies that fail to pair a robust automation strategy with their cloud ambitions will ultimately find the transition to cloud both disappointing and costly.
For far too many enterprises, automation is seen as a way to remove the mundane, cut costs, and move faster. But it is not the mundane that consumes time.
The truth is that operations teams have plenty of tools today to move faster. The reason they don't use them? Networks are notoriously fragile. How else can you explain the draconian change controls we all have experienced? When you are uncertain, you lock it down.
If this is true, then the number one goal of automation needs to be to improve reliability. If networks were reliable, then enterprises would move more quickly. And they would, of course, spend less time on the mundane while also enabling more efficient growth.
Be inclusive: Automation is, of course, about workflows. But if the people identifying workflows are all from the same team, the only workflows that will be identified will be those that exist within that team. What enterprises will find is that the key to unlocking automation is having a bigger picture view. Being inclusive will help surface these workflows.
Inclusion is not just about people. The most effective automation architects will be inclusive, bringing both the virtual and physical worlds together as they automate all the things.
If you can’t see it, you can’t automate it: A surprising number of automation discussions start with the tools required to drive programmatic change and the APIs with which they interact. But the most basic premise of automation is see something, do something.
If you cannot see it, you cannot automate it. The implication here is that automation starts with monitoring. When combined with the previous point about being inclusive, this means that monitoring strategies need to consider more than just a single domain. Seeing the physical or virtual realms alone is insufficient, especially when applications and the workflows that manage them span both.
I used to be a network operator in a past life, and I can tell you the biggest challenges I had were maintaining visibility across diverse operating environments—and that was in a centralized world. As we live our lives closer to the edge, this problem is going to be amplified. Make monitoring a prerequisite to design, not an overlay applied after applications are running and packets are flowing.
Automation must be evergreen: The most important workflows will evolve with your applications. If you set it and forget it, you will find that unlike a fine wine, automation will not age well. Worse, since automation will tend to be built on top of or alongside other automation, as things evolve and become out of date, your operations will suffer from infrastructure rot—the phenomenon where increasing interdependencies left unchecked will add to operational fragility. This is precisely the opposite of designing for reliability.
As machine learning and artificial intelligence continue to take root, this also means leveraging new technology to expand and improve on existing automation frameworks. This means constantly surveying the technology landscape to take advantage of the best that the industry has to offer.
Lock it up: You'd think it's common sense by now, but one thing that still seems to be an afterthought in automating the edge is securing it all. The rise of 5G and IoT services that are pushing the need for distributed edge architectures only increases the need to secure every single scattered end point.
Unfortunately, if you are expecting every device and end-point maker to secure their device—and do it well—then you’re simply delusional. It’s going to be up to the network operator to deploy a high-volume, pervasive way for the network itself to act as the gatekeeper for all the devices connected to it.
You obviously want a network that can identify bad actors and quarantine them, but you also want to be able to take preventative measures to get ahead of it. The network is a source of information and should be positioned in a way that can sense malicious behavior and automatically trigger workflows that remediate it.
Here again, we see the role of visibility. If you cannot see it, you cannot stop it. Invisibility is a hacker’s best friend.
The death of human operators?
As it turns out, the demise of the modern operator has been greatly exaggerated. Companies primarily interested in reducing costs do not hire the talent required to pursue an aggressive automation strategy. They stay with what they know, negotiate harder with their suppliers, and slowly cut their way to progress.
Automation is going to be associated with companies that are growing. You simply cannot grow your operations linearly with your business. Think of automation as a means of growing more efficiently. It is not about removing the operator but rather making the operator more powerful.
Ultimately, machines will prove a great aid in getting more done. But for the foreseeable future, their human partners will provide meaningful context and the conscience required to make the right decisions. Yes, there will be some environments that are more automated, but for the vast majority of enterprises, automation is more a steroid than an operator-ectomy.