AIOps Tackles the Data Conundrum

While large enterprises are the beneficiary of massive volumes of data produced both internally and received from external sources, they can't truly benefit from the information if they lack the capabilities to ingest and analyze it. And, if they analyze only a percentage of the data, enterprises risk making business-critical decisions based on incomplete data sets, which invalidates their findings. According to Forrester, between 60 and 73 percent of data within an enterprise goes unused for analytics.

This problem has only become more acute, to the point that the growth of corporate data, combined with the highly distributed nature of enterprise infrastructures and the siloing of network functions, makes issues on the infrastructure virtually impossible to pinpoint. As a result, many of those issues go unresolved or, in the case of sporadic anomalies, are only temporarily addressed. When that happens, the issue can be patched repeatedly while the root cause remains a mystery.

All these circumstances lead to a backlog in operations, slowing down or crippling the infrastructure, and impeding corporate innovation. The prospect of putting even more data on the infrastructure, which could bring throughput to a near standstill, is out of the question.

AIOps thrives on data

Now things are changing, however. Businesses are turning to artificial intelligence and AIOps (Artificial Intelligence for IT operations) to prevent or resolve high-severity outages and other ITOps problems more quickly on hybrid, highly decentralized, data-driven infrastructures. AIOps doesn’t flinch at a deluge of data – it thrives on data.

But what does AIOps do? An AIOps platform automatically learns the “normal state” of each critical business service and the underlying behavior of the supporting hardware and software services, making it easy to automatically flag anomalies. Using big data analytics, machine learning, and other artificial intelligence technologies, it can then automate the identification and resolution of infrastructure issues ranging from performance or availability issues to an all-out outage of the infrastructure.

While AIOps is nothing new, the application of AIOps on enterprise infrastructures is a recent development that’s been driven by the availability of massive amounts of data from a broad array of sources.

Gartner says a core function of AIOps is “ingesting data from multiple sources, including infrastructure, networks, apps, the cloud or existing monitoring tools (for cross-domain analysis).” Arguably, the growing popularity of AIOps is that it’s the engine for making sense of all that data – where manual correlation and assessment of that data would be futile.

AIOps is best applied proactively and reactively

AIOps is surpassing first-generation technologies in reacting to and solving a number of particularly thorny issues, and forward-looking organizations are deploying a lifecycle approach using AI-powered analytics to continuously optimize operations, and proactively prevent and/or resolve future issues. There’s no room for “later” when the infrastructure is at risk now. These organizations represent the next-generation practitioners of AIOps, who are moving beyond reactive noise reduction.

For example, AIOps today is being applied to workload balancing; it can automate the rebalancing of workloads across servers, preventing an imbalance that could take a toll on the performance of any one of the servers.

A major premise of AIOps is the ability to act in real-time, both on streaming data and historical (stored) data. Real-time capability can prevent crippling downtime, but also provide invaluable real-time data to the AI engine to help it “anticipate” such problems in the future.

As early as 2017, Gartner said that “Consistency in performance of data and analytics platforms is key, and only AIOps technologies are equipped to monitor the modern data and analytics platforms.”

Industry research firm Enterprise Management Associates (EMA) spoke to what AIOps has already achieved.

“While AIOps was not the most pervasive technology associated with advanced IT analytics in our research, it was the most effective and pervasively advanced. Indeed, AIOps showed the highest success rates, the greatest likelihood of supporting DevOps, IoT, and AI bots, and led in use case capabilities as well,” said Dennis Drogseth of EMA.

Researchers and IT practitioners alike speak of the need to analyze the volume, variety, and velocity of data in the enterprise as a defining challenge in an age of unfettered data growth. In that respect, AIOps has taken a prominent position and will need to prove its sustaining value over a larger number and broader variety of use cases. For a quick read on how to launch an AIOps initiative, see How to Get Started with AIOps.