IT infrastructure management software has taken on a distinctly analytical flavor, as enterprise IT pros struggle to keep up with the rapid pace of DevOps and technology change.
Enterprise IT vendors that weren't founded with AIOps pedigrees have added data-driven capabilities to their software in 2018, while startups focused on AI features have turned heads, even among traditional enterprise companies. IT pros disagree on the ultimate extent of AI's IT ops automation takeover. But IT infrastructure management software that taps data analytics for decision-making has replaced tribal knowledge and manual intervention at most companies.
For example, Dolby Laboratories, a sound system manufacturer based in San Francisco, replaced IT monitoring tools from multiple vendors with OpsRamp's data-driven IT ops automation software, even though Dolby is wary of the industry's AIOps buzz. OpsRamp monitors servers and network devices under one interface, and it can automatically discover network configuration information, such as subnets and devices attached to the network.
"You can very easily get a system into the monitoring workflow, whereas a technician with his own separate monitoring system might not take the last step to monitor something, and you have a problem when something goes down," said Thomas Wong, Dolby's senior director of enterprise applications. OpsRamp's monitoring software alerts are based on thresholds, but they also suggest remediation actions.
Dolby's "killer app" for OpsRamp's IT ops automation is to patch servers and network devices, replacing manual procedures that required patches to be downloaded separately and identified by a human as critical, Wong said.
Still, Wong said Dolby will avoid OpsRamp version 5.0 for now, which introduced new AIOps capabilities in June 2018.
"We're staying away from all of that," he said. "I think it's just the buzz right now."
Data infiltrates IT infrastructure management software
While some users remain cautious or even skeptical of AIOps, IT infrastructure management software of every description -- from container orchestration tools to IT monitoring and incident response utilities -- now offer some form of analytics-driven automation. That ubiquity indicates at least some user demand, and IT pros everywhere must grapple with AIOps, as tools they already use add AI and analytics features.
PagerDuty, for example, has concentrated on data analytics and AI additions to its IT incident response software in 2017 and 2018. A new AI feature added in June 2018, Event Intelligence, identifies patterns in human incident remediation behavior and uses those patterns to understand service dependencies and communicate incident response suggestions to operators when new incidents occur.
"The best predictor of what someone will do in the future is what they actually do, not what they think they will do," said Rachel Obstler, vice president of products at PagerDuty, based in San Francisco. "If a person sees five alerts and an hour later selects them together and says, 'Resolve all,' that tells us those things are all related better than looking at the alert payloads or the times they were delivered."
PagerDuty users are intrigued by the new feature, but skeptical about IT ops automation tools' reach into automated incident remediation based on such data.
"I can better understand the impact [of incidents] on our organization, where I need to make investments and why, and I like that it's much more data-driven than it used to be," said Andy Domeier, director of technology operations at SPS Commerce, a communications network for supply chain and logistics businesses based in Minneapolis.
SPS has built webhook integrations between PagerDuty alerts and AWS Lambda functions to attach documentation to each alert, which saves time for teams to search a company wiki for information on how to resolve an alert. This integration also facilitates delivery of recent change information.
"But if I want to do something meaningful in response to alerts, I have to be inside my network," Domeier said. "I don't think PagerDuty would be able to do that kind of thing at scale, because everyone's environment is different."
A diagram of PagerDuty's Event Intelligence process
From IT ops automation to AIOps
AIOps is far from mainstream, but more companies aspire to full data-driven IT ops automation. In TechTarget's 2018 IT Priorities Survey, nearly as many people said they would adopt some form of AI (13.7%) as would embrace DevOps (14.5%). And IT infrastructure management software vendors have wasted no time to serve up AIOps features, as AI and machine learning buzz crests in the market.
Dynatrace's IT monitoring tool performs predictive analytics and issues warnings to IT operators in shops such as Barbri, which offers legal bar review courses in Dallas.
"We just had critical performance issues surface recently that Dynatrace warned us about," said Mark Kaplan, IT director at Barbri, which has used Dynatrace for four years. "We were able to react before our site went down."
[AI and neural networks are] just an evolution of the standard statistics we've always used, and that evolution is much more human than most people believe.
Dennis Curryexecutive director and deputy CTO, Konica Minolta
The monitoring vendor released Dynatrace Artificial Virtual Intelligence System, or DAVIS, an AI-powered digital virtual assistant for IT operators, in early 2017. And Barbri now uses it frequently for IT root-cause analysis and incident response. Barbri will also evaluate Dynatrace log analytics features to possibly replace Splunk.
Kaplan has already grown accustomed to daily reports from DAVIS and wants it to do more, such as add a voice interface similar to Amazon Echo's Alexa and automated incident response.
"We can already get to the point of self-remediation if we make the proper scripts in a convoluted setup," he said. "But we see something smoother coming in the future."
Since Barbri rolled out DAVIS, IT ops pros have embraced a more strategic role as infrastructure architects, rather than put out fires. Nevertheless, enterprises still insist on control. Even as AIOps tools push the boundaries of machine control over other machines, unattended AI remains a distant concept for IT infrastructure management software, if it ever becomes reality.
"No one's talking about letting AI take over completely," Kaplan said. "Then, you end up in a HAL 9000 situation."
The future of AI looks very human
Konica Minolta Inc., a Japanese digital systems manufacturing company, teamed up with AIOps startup ScienceLogic for a new office printer product, called Workplace Hub, which can also deliver IT management services for SMB customers. ScienceLogic's AIOps software will be embedded inside Workplace Hub and used on the back end at Konica Minolta to manage services for customers.
But AI will only be as valuable as the human decisions it enables, said Dennis Curry, executive director and deputy CTO at Konica Minolta. He, too, is skeptical about the idea of AI that functions unattended by humans and instead sees that AI will augment human intelligence both inside and outside of IT environments.
"AI is not a sudden invention -- I worked in the mid-1990s for NATO on AI and neural networks, but there wasn't a digital environment then where it could really flourish, and we have that now," Curry said. "It's just an evolution of the standard statistics we've always used, and that evolution is much more human than most people believe."
Dig Deeper on Systems automation and orchestration
