Getting Started with AIOps

I started my career as a software developer during the first phase of the commercial internet. Back then, we were focused on getting systems online. Access and visibility were our goals, and life was relatively simple.

Because we designed applications linearly — only concerning ourselves with a database, server and web server — it was easy to see the data that we needed to monitor an application's performance.

But times change and data grows.

We've come a long way in building applications more efficiently through the advents of containers, serverless environments and microservices.

The good news is that we've solved for the fact that applications are now accessed outside the data center, whether that be through the cloud, edge computing and soon 5G.

The bad news is that we can't see all the data coming off applications that influence their performance.

Even if we were able to see all the data, we've reached a point where it's impossible for humans to process it in a way to make any meaningful decisions related to application performance.

While we could carry on with a traditional approach to application performance monitoring — one of war rooms and frantic searches for root causes — can we really afford to?

As application data has skyrocketed so too have consumer expectations.

Application loyalty is the new brand loyalty. If users aren't getting a superb experience from a company's applications, they'll find a brand that can deliver the digital experience they crave.

We need a new approach to stand a chance at maintaining customer loyalty, not to mention the sanity of IT operations.

From operations to strategy

AIOps brings artificial intelligence and machine learning to IT operations. Rather than monitoring the health of backend systems independently, now we can assess system health based on a user's interaction with an application.

Here's how it works.

An AIOps platform uses machine learning to take every data point best present inside of an IT ecosystem to figure out normal and acceptable levels of application performance. The platform then alerts us when application performance is at risk of falling below an acceptable level.

Not only do we get alerts, but thanks to AI, we get recommendations on how to address a potential issue based on what's proved successful for solving like issues in the past.

An IT operator can look at a recommendation, confirm that it's the right course of action and initiate a response, for instance rolling back a specific networking policy or re-configuring an IP address.

Plus, we can get these alerts and initiate these actions from our cell phones, which means no more derailed weekends when application performance goes awry.

With AIOps, no longer are we asking predefined questions to gauge application and system health. Instead, we're putting our trust in a more capable processing engine to give us answers to unasked questions.

It's a hard concept to get our heads around, but when we do, we see a world in which IT can operate proactively, not reactively.

But the story extends beyond efficiency. When we take this idea to the next level, it means that every decision an IT practitioner makes is no longer operational — it's strategic.

The user comes first

AIOps challenges us to stop thinking about infrastructure from the inside out. We're no longer prioritizing resources around servers, storage or the network — we move to model where we're prioritizing users based on the individual customer experience.

No human at a large enterprise can look at customer logins to an application and readjust resources according to the buying power of the individual, but this is what we're up against.

I need to know that if my customer is having a difficult experience, my IT infrastructure can rapidly respond, because by the time a traditional alert rolls around, it's too late. Not only have massive amounts of transactional dollars been lost as a result of a database or processing failure, but we've lost potential customers for life.

The next time we make an IT investment, we need to do so with one question at the top of our minds: can we measure this investment against customer experience?

The only way we can answer this is by onboarding AI and machine learning through an AIOps platform, because we can't make these connections on our own.

Taking on the future

AIOps is such a fundamentally different way of considering the role of IT that there's no straight path forward. The best starting point, however, is gaining experience with an AIOps platform.

To that end, we've rolled out an on-demand AppDynamics On-Premise Install Lab.

The lab allows you to test drive AppDynamics, create customized dashboards and understand why Cisco's AI-powered AppDynamics platform is so powerful.

Another key part of the AIOps journey is evaluating your IT organization through the lens of the customer.

Once we can measure the user experience based on the investments we're making in infrastructure, we can then start to envision what our IT organizations can look like.

This will be different for every organization but it will always require executive buy-in and a new level of integration between IT and business leadership.

To begin making progress on this front, I'd encourage you to connect with me about receiving an executive briefing on AIOps. We'll bring experts from different domains together to co-create a broad vision with our customers and then develop tangible ways to leverage AIOps tools to drive IT efficiencies and business value.

When it comes to AIOps and the future, I'd invite you to think about what could happen if we no longer solve for failure but rather prevent it from happening in the first place.

I know this is a challenging thought, but, if we put our minds together, we can do nothing short of shaping IT's evolution.

Learn more about AIOps, application performance monitoring and the result of the WWT and AppDynamics partnership by listening to an episode from our TEC17 podcast.