Dell study finds autonomous operations and AIops adoption lagging

It’s inevitable: Data generation is only going to continue accelerating. 

But while its daily accumulation is moving to near-infinite levels – from petabytes, to zettabytes, to yottabytes (the largest unit approved as a standard size by the International System of Units, equaling one septillion bytes and named for the Greek letter “iota”) – the amount of people in the world to curate and manage data is finite.

Organizations must fundamentally rethink digitization to not only keep up, but to master the process, and data in general. The answer to this: Autonomous operations (AO) and AIops – the process of applying AI to IT operations – according to a new report commissioned by Dell Technologies. 

“We have to be heading in the direction that data is going to get created,” said John Roese, global CTO for products and operations at Dell. “We have to improve the ratio of people to infrastructure.” 

The report, “The State of Autonomous Operations: Why IT Automation is Driving Digital Transformation Success” found that, while organizations are progressing well in their digital transformation, many lag in their adoption of AO and AIops. 

Enthusiasm for AI and AIops strong, but strategy is lacking

The analysis classified 59% of organizations as either “digital adopters” or “digital leaders,” but by contrast, only 15% of organizations exhibit the same level of maturity when it comes to adopting automation. While IT employees are enthusiastic about AI and AIops – often moreso than management teams and other departments – 60% of respondents said their teams didn’t have a strong strategy to implement it, and 77% believed that their organization struggled to create a culture of innovation.

Barriers to adoption included a lack of in-house skill sets and expertise; fragmented, aging, and siloed IT infrastructure; security concerns; and a lack of both budgets and resources. 

Yet at the same time, nine in 10 organizations acknowledged that they are struggling because their IT staff is spending undue time on manual, repetitive tasks that could otherwise be automated. Also as a result of this, more than two-thirds of IT decision makers said their organizations are vulnerable to security threats, and that they struggle with overall speed of IT as well as IT staff retention. 

AI and AIops won’t eliminate jobs

Roese emphasized that just one area, IT monitoring, has been automated by more than half of organizations. Otherwise, few organizations appear to be automating a broad range of IT activities. Some of this is due to reticence on the part of management and non-IT departments, Roese said: There is a general fear that their jobs will become obsolete. 

However, Roese has seen the opposite and points to the automation of security event management monitoring. That process has been almost completely automated and funneled into machine intelligent frameworks. And, in return, organizations “can’t hire enough” analysts to look at upstream distilled information, he said. 

Another example is site reliability engineering. Developed and coined by Google, this practice involves incorporating and applying aspects of software engineering to infrastructure and operations problems. But, as Roese said, automated infrastructures can run into a whole host of problems that require human attention – thus creating a new set of jobs and required expertise. 

He also underscored the benefits that automation can have when it comes to data scrubbing and data cleansing. This has traditionally comprised as much as 80% of data scientists’ work. But “there’s absolutely no reason for that,” he said, adding that heir skills can be applied to more important areas. 

Roese agreed that with any technology implementation, manual human processes go away, and efficiencies and capabilities increase. But people are still required to “program it, manage it, interpret it,” he said. “Now the job is making sure automation works. You have to teach people how to be experts in automation.”

Enterprises get AO AIops

As a positive sign for AO and AIops, the study did reveal that 20% of identified “AO adopter” organizations said there were no barriers to the process. And most other organizations recognize the importance and benefits of automation, with 87% saying they are enthusiastic about it. 

Similarly, 91% of IT decision makers understand that automation is the future: They believe that in 3 to 5 years their IT systems will be able to react autonomously to align with business objectives.

Tech leaders also recognize the benefits automation can have on offering better service, enhancing security, supporting business objectives and growth, and improving IT employees’ work-life balance. Further afield, they said that AO and AIOPs will make them better equipped to onboard and leverage new technologies; deliver new products and services that they “aren’t able to imagine or understand yet”; and operate at speeds that are “currently unimaginable.” 

In short, AO and AIOPs have the potential to dramatically augment IT departments and enable staff to support digitization, Roese said. 

AIops calls for more humans to embrace automation

In general, the most advanced storage systems have a ratio of one administrator for every 10 petabytes of data, he said. But in the case of zettabytes, that ratio increases to the tens of thousands. As such, it’s critical that organizations improve the infrastructure-human ratio – and that by three to four orders of magnitude. A more holistic approach to AO and AIops is necessary, and IT teams must lead the charge in embracing innovation. 

When they do, organizations won’t run into a “human bottleneck,” and they can significantly scale their talent pool. 

“We do have a talent problem,” Roese said. “We have not had enough people in computer science, engineering, and other IT areas for a long time.” 

Still, he emphasized, “the adoption of automation in a vacuum doesn’t mean anything.” Organizations must judiciously capitalize on new capabilities to create a “virtuous cycle of data”: gathering it, studying it and applying the output to learning and reprogramming systems and updating behaviors. 

“You have to evolve with the automation. It is a shift, a journey,” Roese said. 

He added that, when applied correctly, “automation is a really good augment to humanity, and a necessary one.”