How Distributed Cloud Computing Drives IT Automation

Recently, I was the keynote speaker at the AppViewX digital event, “Simplify Application Delivery 2021.” While there were many sub-themes to the event, the one I focused on is automation, as I believe this is the single most important capability for application delivery moving forward.

The application delivery landscape is evolving rapidly and shifting from a vertically integrated hardware stack to a set of cloud native capabilities. While this significantly increases agility, it does raise the bar on complexity, driving the need for automation.

Distributed Clouds: Paradigm Shift in Cloud

I used the first part of my keynote to describe how the rise of distributed computing is changing application delivery. It’s important to understand why distributed clouds are fundamentally different than every other compute model, including traditional cloud computing.

Looking back at on-premises, hosted and cloud computing, while the financial model for these shifted from CAPEX to OPEX, the operating model did not, as they were all based on a centralized compute function. In this case, IT pros would run workloads in a data center or cloud and front-end it with an application delivery controller (ADC). If the location was the business’s own data center, the product of choice was a physical ADC. With cloud, virtual ADCs were used.

With distributed computing, applications are created by accessing workloads or data from public clouds, private clouds, and edge location, giving rise to the concept of composability.

Applications are no longer vertically integrated stacks but rather lightweight, cloud native services that are “composed,” which increases business speed and agility. With distributed clouds, the primary unit of compute evolves from a virtual machine to a container, which is ephemeral in nature. Containers can be spun up, run for a few minutes, and then deprecated just as quickly.

Legacy Application Delivery: Too Slow for Distributed Clouds

The problem with traditional application delivery is that even a virtual ADC can take hours to load – far too long for cloud native systems. This is driving the evolution of ADCs into a number of new form factors, such as a set of containerized services or even API-level ADCs where the functions can be called by an application when needed. Now ADC functions can be spun up when a container might require it.

But how is this to be managed? With cloud native systems and distributed computing, events happen far too fast for people to manage application delivery. This is the role that automation plays, eventually leading to an AIOps model where artificial intelligence is used to make decisions on what changes are needed and when.

For IT pros, it’s important to evolve their thinking around automation from being task-oriented to intent-based. While it’s true, IT automation has existed for some time, the effectiveness of automation frameworks such as a Puppet, Chef and even Python are limited. In this case, these tools are used to automate specific tasks, and this works fine in static systems. Task automation won’t work in highly dynamic environments because the scripts would need to be constantly upgraded.

AI-Based Automation: Necessary for Native Cloud

Automation needs to evolve to an AI-based, closed loop model, where the intent of rules is continuously being analyzed and applied. This enables true zero touch automation as the machines will run the network.

A couple years ago, IT pros often scoffed at the idea of ​​fully autonomous IT operations, but that attitude has changed. I recently ran an AIOps study and found that 97% of respondents would trust AI to run their IT environments and 99% believe AI is important for managing cloud and application performance.

During the panel I did at the event, one of the topics we discussed was using automation to implement zero trust and this is a perfect use case for AIOPs. With zero trust, policies are created to allow specific devices or workloads to connect with others only when explicitly allowed.

Task-based automation would be sufficient in a static environment as the policies could be set up once and then applied. In dynamic environments, such as a distributed cloud, where workloads are constantly being created and then shut down, people could not update the zero trust policies fast enough to comply, but machines can.

IT is at an inflection point where we are evolving from centralized clouds to distributed clouds and this will enable businesses to digitally transform faster than ever. As this happens, IT pros need to embrace closed loop automation for application delivery. This will ensure that the right ADC services are deployed as per business policy, without having to introduce the long lead times created by manual operations.

Leave a Comment