You are here

You are here

31 AIOps deployments in the saddle: Lessons learned

public://pictures/dennis_drogseth.jpg
Dennis Drogseth Vice President of Research, IT Megatrends, Analytics and CMDB Systems, Enterprise Management Associates
 

EMA conducted 31 customer interviews for its recently released AIOps Radar Report and came away with several key findings. Our goal was to confirm what 17 vendors told us in answering our questionnaire about design points, administrative requirements, and outcomes. And this the customers largely did. But the deployment conversations also pointed out what really worked as their adoptions went forward.

The user comments below come from a mix a mix of large enterprises, telecommunications vendors, and MSPs from North America and Europe, with roles ranging from vice president to director to manager.

Here are the most compelling comments, grouped by category. These should provoke a much larger discussion, summarized immediately after the customer comments.

On vendor selection

"Our AIOps platform's support for automation, machine learning. and integration; the relative ease with which we could achieve our required customizations; and the ability to support almost every technology vendor in the end gave us the most bang for the buck" over others evaluated.
Technology and business manager at an MSP

"Our company has grown through several mergers, each with its own diverse tool sets, and we needed our AIOps platform to assimilate what we have and create a common, single pane of glass."
Director of enterprise management at a financial services enterprise

"We purchased our AIOps platform primarily to promote IT-to-business alignment and business outcomes across cloud and existing infrastructures."
Solutions manager at an IT infrastructure and services provider

One overarching theme behind the move to AIOps was th unification of IT through a what many interviewees called a "single pane of glass." This included integrating existing tool sets with added advantages down the road expected from tool consolidation.

Support for integrated automation was also key—most often workflows as well as some type of IT process automation or runbook, along with configuration management capabilities. Ease of deployment was often a priority, along with concern for administrative overhead. Price was rarely top of mind, but it was always a consideration with interviewees.

On a more technical front, some adopters wanted full visibility into their tools' machine-learning capabilities, to understand how they worked and how they might be trained. Others were more interested in solid, quick results without focusing so much on understanding what lay behind the machine-learning heuristics.

And of course data was always a source of interest—events, time series, log files, and configuration data being in the forefront.

On deployment and administration

"It took us about three months to get through deployment where we could see substantial value. Administration on an ongoing basis is fairly easy and doesn't require a lot of engineering. Right now, we have roughly the equivalent of five full-time employees working with our AIOps platform in support of roughly 100 customer environments."
Project manager at a telecommunications company

"Our AIOps platform has turned out to be lightweight to manage, with less than one person dedicated to ongoing administration."
Director of enterprise management at a financial services enterprise

"There are always politics associated with process improvements, so we do get challenged, and our AIOps platform is a huge help in managing the dialogue."
ITSM manager at a financial services enterprise

"Our AIOps platform has become so popular that we've become victims of our own success. There's been a great response to the tool, but of course that puts more pressure on my team to constantly deliver extensions to the solution."
Principal engineer at a communications and entertainment services company

These comments underscore the variety of administrative requirements associated with AIOps, as well as the reality that AIOps can and should change the way IT works—with all of the resulting political benefits and challenges. What EMA has seen in our research consistently is that, given the strategic nature of AIOps deployments, top-down executive involvement is critical to success.

This is true because IT organizations are not only investing in a technology; they are also generally changing how they work, how data is shared, and how IT processes can evolve to reflect more inclusive decision making.

Initially, most deployments focused on anomalies, then slowly worked their way forward toward more predictive and prescriptive values.

On the benefits of AIOps

"The move to AIOps allowed us to unify our operations team with a single-pane-of-glass view and drill down so that we could share information more effectively. In the past, we caught only 3% of our problems proactively. With our AIOps platform, that percentage went up to 88%. Mean time to repair dropped from [multiple] hours to as low as 12 minutes, and we are now able to automate resolutions to known issues."
Director at a global provider of Internet-based entertainment

"One of the values we've seen is minimized change advisory board (CAB) requirements. This has helped tremendously from an OpEx perspective. We no longer need so many people on the call. At the end of the day, they just want to know if a planned change is risky or not."
ITSM manager at a financial services company

"Our stakeholders overall are loving it. They love the ability to consume data in high volumes, slice and dice it as they wish, and drill down or make queries based on their own needs. This includes our DevOps teams, who are able to consume the data at a quick pace and almost instantaneously make adjustments to the development process thanks to our AIOps platform."
Solutions architect at a government agency on the Pacific Rim

"With our AIOps platform, we've been able to consolidate our tool set investments across all tiers of service delivery, with both operational and CapEx savings. Automation is driving new levels of efficiency so we can refocus around higher-value activities. For instance, we've been able to reduce incident volumes by 76% and reduce MTTR for major incidents by nearly 400%."
Director at a global communications service provider

While the benefits users cited above center on core values such as tool set consolidation, mean time to repair (MTTR), and incident reduction, they also highlight improvements in change management and DevOps.

One of the most interesting business benefits not cited above was the idea of integrating weather variables into other business key performance indicators, since weather turned out to impact business process effectiveness for at least one user organization. Indeed, benefits listed in our research spanned 27 different areas. Some of the more prevalent beyond incident reduction and improved MTTR included:

  • Event noise reduction
  • Improved OpEx efficiencies across IT
  • Real-time insights and historical trends about IT services
  • Proactive ability to prevent problems
  • Less time spent writing and maintaining rules and thresholds
  • More effective migration to public cloud
  • Better correlation between change and performance
  • Faster time to deliver new IT services
  • Better alignment of IT services with business performance
  • Accelerated capabilities for digital transformation

AIOps is the catalyst

These are just a few highlights from what we learned, as each deployment offered something new. The notion that IT issues are generic obscures the fact that IT organizations are as diverse in their DNA as human beings are, with distinct business requirements, as well as cultural, political, and technological attributes.

AIOps deployments emerge as a catalyst in promoting more progressive and more mature IT organizations, and should be approached as such.

For more details on the lessons learned from 31 AIOps deployments, read EMA's AIOps radar report

Keep learning