Artificial intelligence started out as an experimental part of organizations. But considering AI spending hit $35 billion last year, it has to get past the pilot programs and into cross-company strategies. One way AI will start adding value is by getting helping out IT operations, or AIOps. AIOps is the hot new buzzword that acts as a generalizing umbrella term for all things big data analytics, machine learning and AI when applied to identifying and resolving enterprise tech issues and streamlining operations.
Shreyans Parekh, senior manager of corporate strategy and go to market at AppDynamics application and infrastructure monitoring, predicts this will be the year when we will see AIOps make a name for itself in the future of enterprise operational efficiency, reliability and security.
“It’s certainly a trending hashtag, but AIOps is the advent of ML and AI algorithms — very complex algorithms — that are now able to make the job of IT operators and developers much more efficient and easier, so that the IT operators and developers can then shift their jobs and workstreams to actual business and revenue earning potential,” Parekh said.
He says AI and machine learning will do this through smarter anomaly detection, even before hitting the site reliability engineer’s error or warning threshold.
Parekh asked, “How do you investigate in a really wide, very complex distributed system? How do you investigate pinpoint errors and able to troubleshoot and remediate very quickly? And the way that we’re able to do that is through our AI and ML engine, which is called Cognition Engine.”
The AppDynamics Cognition Engine is one such example of a collection of machine learning algorithms that not only automate anomaly detection and root cause analysis, but they allow for faster mean time to resolution (MTTR) and “intelligent” issue alerts. Parekh says it also takes up little bandwidth and the agent is quick to install locally, out of the box. He says it takes a few weeks to get up and running across an enterprise.
The tooling also offers threshold recommendations based on specific kinds of business transactions and distributed systems, or you can manually set them.
One example of AI-backed serverless is seasonal provisioning of infrastructure in industries like retail, travel, and banking. This can automatically expand hosting coverage during holiday rushes, and save money by reeling it back during AI-predictably slow times.
Just like DevOps looks to reduce silos and operations complexity, Parekh says that AI-backed serverless allows teams to create greater efficiency, increasing team and individual potential, while building trust and transparency with the tech side. Observability is going to become part of the AIOps umbrella, which should allow for the increased shared visibility that comes from open trace and open metrics.
He argues that the core of AIOps isn’t the technology but increasing team efficiency and driving more positive business outcomes. Organizations are still spending time trying to manually remediate severity-one issues. Parekh says AIOps tolling like Cognition Engine will not only quickly identify and resolve these Sev1 issues, but it will free up time for SREs to focus on more revenue-driving initiatives.
AppDynamics is a sponsor of The New Stack.