AIOps-enabled ITOM for Hybrid Infrastructure

IT operations management software (ITOM) has gone through several iterations in the past decade as infrastructure technologies have evolved, and customer requirements have become more multifaceted and complex. IT operations first needed to deal with the advent of public cloud, and then multi-layered requirements as enterprises gravitated to hybrid IT environments comprising internal data centers and multiple public clouds. During this shift came the advent of IoT and edge computing, real-time analytics needs and shadow computing – all placing a tremendous strain on internal IT to support and troubleshoot.

IT operations teams recognize the need for better efficiency and have looked to automation tools to reduce the strain on human operators. Classifying and routing tickets, integrating tools and data sets for a central dashboard view and automating routine configurations and patches have been common initiatives of the past several years.

Artificial intelligence for IT operations or AIOps, is the latest phase, focusing on algorithms that learn over time to improve suggestions for IT incidents and filter alerts streaming constantly from virtual, cloud and on-premise systems. Despite a growing number of large and small vendors developing AIOps applications, we are still in the early phases of this sector with many questions still to be answered.

On balance, however, IT directors seem positive;The OpsRamp State of AIOps Report, 2019, found that 87% of technology pros agree that AIOps tools are delivering value through improved hybrid infrastructure resilience, data-driven collaboration, and proactive IT operations. IT organizations are using these tools primarily for intelligent alerting, root cause analysis and anomaly detection. Survey respondents cited three major advantages of AIOps: better productivity from eliminating low-value tasks (85%), rapid issue remediation with faster root cause analysis (80%), and better infrastructure performance through reduction in incident and ticket volumes (77%).

Where AIOps goes wrong

People can only trust machines to a certain point. And to be honest, AI hasn’t exactly had the best run – recall the many privacy issues which appeared with consumer products such as Amazon Alexa & Echo, Google Home and others. In the world of IT operations (and business generally speaking), the issue is about trust and accuracy. A survey of CEOs from KPMG International revealed that 67% of those leaders ignored computer-driven data analysis because those findings contradicted personal intuition or experience. These concerns have driven changes in some vendor products to include AI "explainers" that deliver in-app visualizations showing how the algorithms work and reached their conclusions.

Another problem is coming to agreement on what the true value to IT operations is, from AIOps. Is filtering and correlating alerts to reduce manual effort and point to possible root causes enough to justify an expensive software product if the algorithms are still immature? Training an AI model on "normal" versus "abnormal" scenarios is a moving target ripe with inaccuracies. Something abnormal may not be an actual problem but could still cause IT Ops teams to chase it down if the AI system says to do so.

New technology always has glitches and learning curves. Using old methods to manage altogether new and constantly changing environments, however, is unlikely to prevail for much longer. Says Bhanu Singh, Senior VP, Product Development and Cloud Ops at OpsRamp in The New Stack: "Just like DevOps drove the cultural shift in the organization, AIOps is empowering a data-driven organization to uncover holistic insights from connected and disparate data to drive decision-automation."

OpsRamp provides a great information about Aws Cloud Discovery and Management. For more details, visit: www.opsramp.com!