Report: IT Departments Can Benefit from AI, But at High Cost

IT could use AI but can they afford it? That's the question a survey report, "Top Trends in AIOps Adoption: The Future of Digital Operations Management" by OpsRamp Inc., attempts to answer.

The San Jose, Calif.-based AIOps platform provider defines AIOps as "artificial intelligence for IT operations." One focus for AI in IT would be helping to manage the alerts IT professionals receive throughout the day as systems reach thresholds or are nearing failure points or failing. The volume of alerts can become a problem itself if the IT staff becomes overwhelmed by them.

"IT professionals are now drowning in 'alert storms' that negatively impact service availability and increase resolution time for IT outages," the report states. Digital transformation has surpassed the ability of traditional monitoring tools to keep up with the volume, it argues.

"While traditional IT monitoring tools can handle routine infrastructure operations, modern digital operations management requires tools built on a solid foundation of artificial intelligence and machine learning," the report states.

The report cites Gartner's Market Guide for AIOps Platforms, which states, "AIOps platforms combine big data and machine learning [ML] functionality to support IT operations. I&O [Infrastructure & Operations] leaders should begin the deployment of such platforms to enhance performance monitoring now, but plan extension to service desk and automation over the next five years."

OpsRamp also points to a report by Illinois-based market research firm MarketsandMarkets that projects major growth in AIOps: "The AIOps platform market size is expected to grow from USD 1.73 billion in 2017 to USD 11.02 billion by 2023."

The OpsRamp survey of 120 IT executives in the United States found that while they recognize the need for AI and ML to help cope with the mushrooming alerts and tasks being generated by the digital economy, cost and complexity are a barrier to AIOps in the early adoption curve.

"While AIOps adoption is picking up at a fast pace," the OpsRamp report states, "IT teams have a growing set of apprehensions. 54 percent of respondents worry about the accuracy of AIOps prediction models, 52 percent about the quality of large datasets for building machine learning models while 48 percent fret about the IT talent required for supervising machine learning algorithms."

When asked what was holding them back from AIOps adoption, 52 percent of those surveyed cited cost, while complexity concerned 29 percent and another 29 percent worried about "technological limitations."

OpsRamp does not offer easy answers to the questions facing early adopters. As with other AI and ML technologies, lack of skills is an issue. "IT teams will need to hire qualified data scientists to implement AIOps tools," the vendor's report states. "Also, don't forget to train existing staff so that you can finally close the skills gap."

It appears AI training is in the future for IT professionals looking to work in the AIOps world. How practical it will be for IT departments to hire those currently scarce data scientists is a question that needs to be answered. Beyond the price of purchasing an AIOps platform, the cost of training and/or hiring AI experts will be a consideration.

Despite these reservations, the IT executives in the survey apparently see the value in applying AI and ML to cope with the stress the digital economy is putting on traditional IT operations. Automating routine functions scored high, with 74 percent of those surveyed seeing that as one of the two biggest benefits to AIOps adoption. Avoiding service disruption with faster mean time to recovery (MTTR) came in second with 67 percent citing it. Detecting anomalies "by predicting shifts in normal system behavior in dynamic production environments" came in third, with 58 percent seeing that as a benefit.

However, with these benefits, the key to adoption appears to depend on vendors' ability to provide IT departments with technology that is not difficult to deploy and maintain. According to the survey, 63 percent "want an AIOps platform that's easy to deploy, configure, and maintain across their modern IT infrastructure landscape." That is a tall order for a technology still in the early stages of development and adoption.

Predicting that ML has no limits when it comes to managing the future of IT, the OpsRamp report concludes: "The potential of AIOps is still in flux. As AIOps tools grow in sophistication, enterprises expect to save time and money with actionable event context and data-driven recommendations."


comments powered by Disqus

Subscribe on YouTube