The Rise and Future of AIOps: How ML can Resolve Multicloud Complexity

“CIOs today are caught in between a rock and a hard place. The rock is maintaining high availability while doing that as cheaply as possible. The hard place is enabling digital transformation and increasing business and developer velocity,” answered Assaf Resnick, CEO and Co-Founder of BigPanda, at the opening session of RESOLVE ’22.

AIOps versus Multicloud Complexity

Every CIO feels this paradox and tradeoff daily. One long major incident can disrupt a CIO’s transformation agenda, particularly when many business leaders believe that apps running in the cloud should deliver 100 percent reliability. Incident management is a challenge for many IT organizations, and in StarCIO’s recent research, 70 percent of respondents said it typically takes over three hours to resolve high priority (P1) issues. Those issues get more complex – not less – when an organization runs hybrid/multi clouds. So CIOs are looking for answers.

“The rise of AIOps” was the theme at Resolve ’22, a community event for IT Ops, NOC, DevOps, and SRE professionals. I attended the conference as I’ve been writing and reviewing AIOps for several years, and I looked forward to hearing how IT Ops professionals were transforming.

While addressing this paradox is a challenge, Sanjay Poonen, former COO at VMWare and President at SAP, provided advice on solving the operational versus transformation paradox. He says, “The best CIOs understand how software is being developed and model their organization after not just a startup, but the largest company they can find that’s operating like a startup.”

Many CIOs understand this charter, and their digital transformations aim to modernize their business with improved customer experiences, real-time analytics capabilities, and efficiencies through workflow automation.

Translating this directive to IT Operations, operating as an enterprise startup requires them to 

  • Provide higher reliability and performance that exceeds end-user expectations while operating in more complex multicloud and edge environments
  • Improve the mean time to recover (MTTR) from incidents and perform accurate root cause analysis despite the “astronomical data volumes,” as one session panelist remarked 
  • Focus IT Ops on higher-value work by creating ITSM automations that can also reduce errors and eliminate shoulder tapping as an acceptable form of communications 

The Rise of AIOps

Panelists in the “AIOps 2022-2025: Expert predictions” keynote panel shared insights on why IT Ops needs AIOps. James Maguire, Editor in Chief at eWeek, shared today’s reality in simple terms, stating, “Today’s IT systems are too complex for humans to run them.”

It’s not just about going from the data center to the cloud, multicloud, and edge computing. Over the last decade, IT shifted from three-tiered to microservices architectures, virtual servers to serverless computing, and behind-the-firewall software to ecosystems of integrated SaaS, low-code, and customer-facing applications.

Now, ask IT Ops which service is the root cause of the outage when dozens to hundreds of observable systems are pumping out telemetric data. The answer is rarely easy to get to, and while teams are searching for the answers, the incident persists.

Carlos Casanova, Principal Analyst at Forrester, explained the pressure everyone in IT feels daily. He said, “If that system goes down, someone’s on social media complaining about it within a half a second. The public relations fallout can have a greater financial impact than the system being down.”

IT can’t solve these issues today by hiring more people. As Eric Noeth, Partner at Advent, explains, “There’s an incredible tailwind around complexity, and it’s also against a market backdrop of the real scarcity of talent.” 

AIOps is a Strategy that Transforms IT Operations

The rise of AIOps closes this gap by centralizing IT data, providing machine learning capabilities to improve ITSM and IT Ops, and enabling automations across IT’s ecosystem of tools and technologies.

I was on the panel “How AIOps works in the real world” with Sean McDermott, CEO of Windward Consulting, and he shared one of his secrets on implementing successful AIOps programs. He said, “We see AIOps as a strategy, not a tool. AI and machine learning entering into the operations domain is transformational. The use cases we’re focusing on now around event management, correlation, and root cause analysis are just the beginning of the journey.”

Many organizations start AIOps with a POC in a single domain and address several primary problems, such as reducing MTTR in a mission-critical business process. We described how setting up centers of excellence and building product management disciplines can help IT leaders expand AIOps from a single domain and use case to a platform leveraged across the enterprise.

The Future of AIOps

Since we’re only at the beginning of the AIOps journey, it begs us to forecast what capabilities might become available in the near future. Panelists on the “AIOps 2022-2025: Expert predictions” keynote panel had their answers.

Michael Yamnitsky, Managing Director at Insight Partners, believes AIOps platforms will “shift to domain-specific insights” and beyond incident management, root cause analysis, and automation capabilities. He said, “We will see AIOps tools deliver more insights to the DevOps and engineering teams that are actually responsible for fixing the problems.”

Maguire adds, “We need to be able to predict the problems before they happen. We can’t just respond to them anymore and need predictive analytics.”

These are promising capabilities and possibly the closest thing to an “easy button” for IT Ops facing greater business pressures and technology complexities. The rise of AIOps has started in many organizations such as Sony PlayStation, Cardinal Health, Wells Fargo, and Honeywell, and listening to their stories, you can see how AIOps is truly transforming IT operations.

BigPanda sponsored this conference and content.

The views and opinions expressed herein are those of the author and do not necessarily represent the views and opinions of BigPanda.

No comments:

Post a Comment

Comments on this blog are moderated and we do not accept comments that have links to other websites.

Share