In APMdigest's exclusive interview, Colin Fletcher, Research Director at Gartner, talks about Algorithmic IT Operations (AIOps) and how it will impact ITOA (IT Operations Analytics) and APM (Application Performance Management).
APM: For the readers who are unfamiliar, what is AIOps?
CF: Algorithmic IT operations (AIOps) platforms utilize big data, modern machine learning and other advanced analytics technologies to directly and indirectly enhance all primary IT operations functions with proactive, personal and dynamic insight. AIOps platforms enable the concurrent use of multiple data sources, data collection methods, analytical technologies (real-time and deep) and presentation technologies. AIOps platforms represent the evolving and expanded use of technologies previously categorized as IT operations analytics (ITOA).
APM: What advantages can IT Ops gain from AIOps?
CF: I find it is really useful when looking at how you or your operations team can take advantage of predictive, machine learning-enhanced tools to think in terms of how they assist and/or augment your current capabilities. The ideal state or ultimate goal of an AIOps investment is a platform that is capable of continuously, proactively generating insights that are used in support any number of internal and external customers. While AIOps has tremendous potential to deliver on use cases that stretch well beyond core IT operations functions, to date we've seen enterprises get real, tangible value using AIOps platforms to:
■ Make the holy grail vision of a "single pane of glass" a reality across multiple technology stacks and generations most often in support of root cause analysis
■ Rapidly support new digital business initiatives and their accompanying use of the latest disruptive technologies (containers, microservices, IoT, etc.) at scale
■ Achieve the long sought after goal of automated, sustainable, scalable, and most importantly, useful event correlation that works to reduce alert noise/fatigue and speed diagnosis
APM: How does AIOps enable you to get more from your existing data?
CF: To put it as simply as possible, AIOps provides for many a more practical way to get multiple data sources into one platform and apply multiple analytical technologies to that data in an automated fashion to discover the relationships and patterns that lie undiscovered in previously isolated data. This is particularly true in the case of utilizing IT operational data in combination with data generated by applications or infrastructure normally outside of IT's operational visibility. To be clear, this is not to say that this has somehow not been possible previously, of course it has been, but in most cases prior to AIOps, it was cost prohibitive or technically challenging to do so.
APM: Does AIOPs augment or support APM?
CF: Currently AIOps is typically used to supplement APM use cases and/or tooling by providing a much more practical and in some cases cost effective means of filling in the gap between what data is being collected directly by the APM tool and the rest of the supporting applications, infrastructure, security, service, customer/business operational, and configuration data that is rarely directly integrated or utilized in APM tools. This is due in large part to AIOps' emphasis on providing the ability to continuously deliver insights from multiple data sources regardless of the mechanism used to collect the data.
APM: Does AIOps support DevOps?
CF: DevOps teams and particularly application developers gravitate to AIOps tools naturally in their search for data-driven (as opposed to instrumentation-driven) insight to their particular application's behavior.
Many DevOps teams also cite the "democratic" or "agnostic" or "open" nature of AIOps tools that from day one are assumed to be integrated with "something or multiple somethings" to deliver combinative value as well as their genuine support of experimentation and creative use of data for purposes beyond problem solving as reasons for using AIOps tools.
Most frequently we see DevOps teams using AIOps to monitor application and infrastructure performance, troubleshoot issues, and provide dashboards and reporting across entire toolchains that consist of multiple tools used in both development (CI, Test, ARA) and operations (monitoring, CD/Release/Configuration).
