Recently, I had the opportunity to moderate a webinar, Using AIOps to focus I&O resources on what really matters, with guest speaker, Forrester Senior Analyst Rich Lane and Dynatrace SVP of product Steve Tack. In the webinar we covered a range of topics including the types of challenges AIOps can help address, different approaches for AIOps, advice for organizations implementing AIOps, and more.
Because we only had time to scratch the surface of a topic that is a high priority for many organizations, we asked Rich five follow-up questions to go deeper.
- What are the most important insights that teams can get from AIOps? Is it mostly about identifying and resolving problems, or are there other types of answers that are equally or even more important?
Rich Lane: Well, there are a lot of advantages to bringing in as many data sources as possible into one data lake. Even data considered as ‘non-IT’ oriented can help to enrich the overall picture of how the business is functioning or how satisfied digital users are with a particular service they use. Yes, finding application or infrastructure component issues as fast as possible is key in reduction of MTTR, but there’s so much more when you frame it in the context of how is a service performing as a whole. Looking end-to-end across an entire technology stack can bubble up performance or usability issues with these services that IT Operations or development teams were unaware of because they have been looking at each component from a siloed view.
- In the webinar you spoke about the importance of connecting to business outcomes. What are some steps IT ops can take to provide more value to business stakeholders?
Rich Lane: In the digital age where so much of an enterprise’s revenue and customer engagement comes from digital interaction, IT needs to shift from measuring success in terms of server up-time or number of incidents to business outcomes (e.g. Did our latest marketing push achieve a greater number of sales conversions? Does the latest release of our mobile app drive greater customer engagement?)
- How can AIOps help improve collaboration both within IT ops and also across biz, dev and ops teams?
Rich Lane: One of the great things about AIOps from an IT Operations perspective is that it removes finger pointing and the necessity of the outdated ‘war room’ concept. Given that now we can apply machine learning algorithms across large sets of seasonal data to perform complex correlations in real-time, we are getting to root cause immediately. Then, through automation we are engaging only the teams we need to remediate a particular issue. Combine in the ability to use collaboration tools (Slack, Teams, etc.) natively within AIOps platforms, the result is that all the stakeholders, be it operations, sales, development, or marketing, are engaged and looking at the same dataset to make informed decisions.
- You talked in the webinar about some best practices for getting started with AIOps, but what about mistakes to avoid? Any lessons learned from what you’ve seen others do wrong?
Rich Lane: The biggest issue I see is enterprises trying to build a like-for-like system based on how they have done things in the past. This is generally because they have not spent the upfront time going through a review of their processes current state and how an AIOps enabled platform will change those processes. If you think of an incident management workflow, many of those processes as they were written even five years ago can now be automated, which helps teams get back to doing higher-level work.
- Tool rationalization and simplifying the portfolio of monitoring tools are goals that many enterprises seem to have today. What are reasonable expectations for what is achievable in terms of reducing the number of tools with the help of AIOps?
Rich Lane: This is an interesting one as the answer will be different for every organization depending on the approach they take. That being said, there will always be an ecosystem of tools that enterprises will have. There are point solutions (e.g. database analysis, deep network inspection) that will need to be kept but the data they collect can be directed into the AIOps platform to enrich the overall picture of service health. Over time as the AIOps use cases and successes become apparent, the question of, “Do I still need this seemingly redundant tool in this new architecture?” gets answered.