AIOps tools beef up insights, but long-term scope unclear

AIOps tools present a way to cope with IT infrastructure sprawl and complexity, but how far they can go with hands-off automation features is still a topic for debate.

As enterprise IT shops put AIOps tools through their paces, they are divided about just how AI-driven the future of IT infrastructure management will be.

Some early adopters of AIOps tools already have taken a significant portion of traditional IT ops tasks out of the hands of humans. Others are skeptical about the term AIOps itself and see little benefit to advanced IT analytics beyond alert noise reduction. In either case, the technology still has far to go to fully mature.

"AIOps is a very new idea, and not many organizations are using it in anger as yet," said Clive Longbottom, an independent analyst and a TechTarget contributor. "The position most are in are using a mix of tools with a degree of machine learning and lots of rule-based engines, which is the nub of the problem, as the rule-based engine will do what it is told time after time after time, even if the rule is wrong."

A tale of two AIOps tools

AIOps users agree data analytics tools are necessary to address how quickly IT infrastructure has grown and how complex it has become. There, however, similarities between how much they trust AIOps automation may end.

At Carousel Industries, the OpsRamp AIOps tool has reduced 96% of the noise generated by hundreds of thousands of IT monitoring alerts on millions of devices the managed service provider operates for clients. Carousel has also begun to trust OpsRamp's automated service ticket generation and resolution functions.

Tim Hebert, chief managed services officer, Carousel IndustriesTim Hebert

"It has a really robust event correlation engine," said Tim Hebert, chief managed services officer at Carousel, based in Exeter, R.I. "The amount of human effort that we've had to put into getting the job done has been reduced drastically, and that has allowed us to grow our business without adding more human resources."

As Carousel grows more comfortable with AIOps tools, the data they gather and the consistency of their analysis has turned out to be more reliable, in many cases, than manual systems oversight by humans.

"Where we do have an SLA [service-level agreement] breach, it's usually because of a human and not a machine," Hebert said.

However, there's still plenty of room for improvement in the OpsRamp tool, much of it addressed by its summer 2019 release rolled out in June 2019. The tool still must mature in its ability to make inferences between multiple silos of IT, such as networks, servers and data centers, Hebert said.

AIOps automation could take over some 50% of Carousel's repetitive manual tasks, he estimated, once the company tests out new service and topology maps, which link IT service maps with network topology data and manage IT events based on business and operational priorities.

Hebert also said he'd like to see better out-of-the-box reporting features for business stakeholders in future OpsRamp releases, beyond the fact that OpsRamp allows its data to be exported for consumption by third-party business intelligence (BI) tools.

"Most CIOs don't know if a report about a server or system being down is good or bad," Hebert said. "We need to deliver guidance on the business impact of IT issues, not just what it means to the IT organization."

At KeyBank, however, AIOps tools play a much narrower role in IT service management, and humans still decipher actionable insights from their analysis using separate IT monitoring and BI tools. Moogsoft's AIOps tool helps KeyBank narrow down what issues really need a response amid a flood of IT monitoring data, but root-cause analysis remains human-driven.

"What [AIOps tools] are really good at doing is integrating into systems like ServiceNow and looking at the history of how you resolved a ticket the previous time," said Mick Miller, senior DevOps architect at the Cleveland-based financial services company. "Moogsoft can head you down the right track ... but as far as root cause, I'm not trying to get them to do that."

AIOps tools add context, reporting features

What [AIOps tools] are really good at doing is integrating into systems like ServiceNow and looking at the history of how you resolved a ticket the previous time ... but as far as root cause, I'm not trying to get them to do that.
Mick MillerSenior DevOps architect, KeyBank

In addition to service and topology maps, OpsRamp's summer 2019 release includes cross-site connection topology maps that iincorporate routing-layer relationships across WAN links for multi-cloud management. The release also deepens the tool's integrations with AWS, Kubernetes, Mesosphere and Azure Stack infrastructure, as well as open source applications, such as Couchbase, Apache CouchDB and the Elastic Stack.

Moogsoft's AIOps 7.2 release in May 2019 also added connectors to New Relic Insights and Microsoft Teams, as well as updated integrations to tools such as Zenoss and vCenter. Version 7.2's Individual Statistics view will come in handy to KeyBank, Miller said, to provide a more detailed picture of the performance of network operations center admins as they address incidents.

Elsewhere in the AIOps tools market, BMC TrueSight recently expanded the actionable insights and reports it offers with Business Service views for business stakeholders and Event-Driven Compliance for CloudOps, which automates policy-based governance of security whenever a change is made to IT infrastructure.

Still, these updates are incremental steps toward the kind of self-driving, predictive IT infrastructure automation some AIOps tools users dream of, Longbottom said.

"The direction we need to be moving to is idempotency: the capability to tell a system what the desired outcome is and let it sort out the steps that are required to make that so," he said. "I expect such technology to move to the AIOps space and make life easier for system admins -- as long as they do actually know what the desired outcome is."

Dig Deeper on IT systems management and monitoring

Software Quality
App Architecture
Cloud Computing
Data Center