The Five Stages of AIOps Adoption

    It has been said that DevOps is a journey. The same certainly holds true for AIOps. Like DevOps, AIOps is not something you simply turn on or “do” overnight. Instead, it’s a strategy that you embrace incrementally and scale up as you learn to leverage AIOps in newer and more sophisticated ways.

    Recognizing the gradual nature of AIOps adoption is critical for its success. With that need in mind, this blog post discusses the five common stages that your team can expect to pass through as you progress along your AIOps journey.

    AIOps Stage 1: Early Adoption

    The first stage of the AIOps journey starts when you deploy an AIOps-powered platform for the first time. At this stage, you can expect the platform to be:

    • Driven by a limited data set. For example, the platform might ingest logs from a single application that it is helping to monitor. It’s not until later stages that you will move toward larger, unified sets of data.
    • Focused on a specific need or pain point. Rather than helping to manage every aspect of a application’s performance, for example, the AIOps platform might manage just resource allocation.
    • Designed to make recommendations rather than to take actions. The platform will suggest actions that your team can take to improve the performance of your application, but it will not take those actions directly.

    This is the type of deployment that helps teams become familiar with the fundamentals of AIOps and learn how to start integrating it into their workflows. However, because AIOps platforms are limited in focus at this stage and teams do not yet trust them enough to act on their own, they serve only as supplements to IT operations rather than as the core of any workflows.

    AIOps Stage 2: Widening the Scope

    Teams reach the second stage of AIOps when they begin using their AIOps-powered platform on a larger scale. Typically, they won’t add more platforms yet; they’ll stick with the one they originally deployed. But instead of using it to serve just a single purpose (like managing resource allocation, to use the example from the previous section), they will expand its usage to cover wider ground (like managing multiple applications and supporting services as well).

    The team may also take some halting steps toward a more unified data set to power the AIOps platform. Instead of relying on just a single application’s logs, they might incorporate logs from all applications of the same type. The data set would still be small and hardly comprehensive, but it would be broader than the one used in the first stage of AIOps adoption.

    At this point, the platform will continue to make suggestions rather than take automated action. Your team still won’t trust it enough to let it modify anything on its own.

    AIOps Stage 3: Diving into Complexity

    As your teams become more and more aware of the insights that AIOps can provide, they will become eager to receive guidance on more complex issues from their AIOps platform. For instance, rather than looking for recommendations about the resources to allocate to an application, they may want to use AIOps to help understand how performance issues correlate to resource allocations in order to get to the root cause of problems faster.

    AIOps platforms are well suited to parse complex issues like this, but only if they have enough data to understand them. For that reason, stage three of AIOps usually involves further expansion of both the volume and the diversity of data fed to the AIOps platform. Instead of letting your platform analyze only logs from similar applications, for example, you might let it ingest logs from every application in your data center, even for those that are architected in unique ways. Having a large, disparate set of data will help the AIOps platform gain better visibility into how each application is operating and how its performance compares to that of the rest of the data center.

    Still, the team will rely on the platform mostly to make recommendations about how to solve complex problems, not to solve them directly at this point.

    AIOps Stage 4: Playbook-Driven Automated Remediation

    When that changes and you begin embracing the automated remediation features of AIOps for the first time, you reach stage four. At this point, the team finally becomes comfortable letting an AIOps-powered platform sit in the driver’s seat (at least some of the time) and automatically take action to resolve problems that it identifies, like an under-provisioned application.

    The major caveat here, however, is that the platform’s actions will be based on preconfigured “playbooks” that determine which interventions it can make in response to certain conditions. In this sense, the platform won’t act in a truly autonomous fashion; it will simply determine when preset conditions are met, then take action accordingly.

    AIOps Stage 5: Autonomous Remediation

    When you move beyond the playbooks and let your AIOps platform devise its own remediation strategies, you’ve reached stage five. This is when the power of AIOps becomes most visible, because AIOps can finally replace (rather than supplement) the human operators within an IT workflow.

    This isn’t to say that playbooks disappear from the picture. In most cases, AIOps platforms will be planning remediations based on playbooks that they used previously, although they may change some of the steps based on insights they have gleaned from previous operations. The playbook may say that a certain type of application error should be fixed by changing memory allocation by a certain amount, for example. But the AIOps platform may determine that the optimal memory reallocation is actually less than what the playbook specifies and act accordingly.

    By this point, your team has placed its full faith in AIOps. Human engineers will still need to address truly complex issues that AIOps can’t resolve on its own. But for routine workflows, they trust AIOps to handle everything for them. They will also be feeding as much data as they can collect into their AIOps platforms, realizing that the more data they have driving AIOps, the better.

    Conclusion: The AIOps Journey Never Ends

    Stage five isn’t really the end of your AIOps adoption journey, of course. There will always be opportunities to deploy new AIOps platforms or apply AIOps to new workflows. Thus, just as DevOps is a never-ending journey, the AIOps journey is never truly over. Stage five just means you’ve learned to take full advantage of the AIOps platform.

    It’s worth noting, too, that your mileage will of course vary. Every organization’s approach to AIOps adoption is unique. But they all involve incremental expansion and improvement upon the ways in which AIOps is leveraged. Keeping the progressive nature of AIOps in mind is essential for getting the most out of it and ensuring that you don’t stop before you allow your AIOps initiative to realize its full potential.