Computer infrastructure has become more complex and dispersed, resulting in a flood of alerts in IT operations centers. As the volumes grow, MSPs and enterprises can become overwhelmed and challenged by troubleshooting routine problems.
Enter artificial intelligence for IT operations (AIOps), which applies AI and machine learning software to solve the problem by automating and streamlining significant parts of the management process.
The modern data center often looks like a NASA spacecraft control room with large wall panels and teams of data center technicians banging away at keyboards to glean how the computer infrastructure is performing and why. Such vital deductions have become more difficult to discern.
"Today, businesses deal with so much data: multilayered systems, complex applications, microservices, cloud, software-defined network infrastructure and IoT instrumentation," explained Sean McDermott, president and CEO at Windward Consulting Group. "Operations staff do not have the capacity to manage it as they did in the past, mainly with human decision-making. They need machines to help them crunch the numbers."
Why AIOps is gaining interest
Organizations use AIOps platforms to simplify and improve decision-making by contextualizing large quantities of operational data, according to Gartner's Market Guide for AIOps Platforms. Rather than endless streams of information flowing across the computer screen, AI tools associate data. For example, the tools can automatically link appropriate permissions to a person accessing the enterprise network. These connections provide needed visibility into system usage and performance.
Organizations and channel partners must sort through the realized potential and hype of the AI market.
"AIOps is a noisy space, one where certain vendors affix the term to their products, whether or not it seems appropriate," McDermott stated.
While the potential is vast, actual AIOps implementations are currently quite limited.
"The AIOps industry now is in an early stage," said Tanner Bechtel, global director of AIOps, automation and orchestration at services provider World Wide Technology (WWT). "Some businesses have gone headfirst and done advanced work with it, but the system ecosystem is still being built."
Organizations must have a strong ecosystem to construct an AI model because it is a complex process. According to Gartner, the model includes five stages:
- Ingestion. Collect, index and normalize event and telemetry data from multiple domains.
- Topology. Build software that visually displays activity within the computer infrastructure.
- Correlation. Connect events across domains and sources, ideally without much human intervention.
- Recognition. Understand the current state of enterprise infrastructure performance and how it will change.
- Remediation. Train AI tools to take action or make recommendations when problems arise.
New possibilities for MSPs
AIOps offers MSPs and channel partners opportunities to help large enterprises build their own AIOps solutions. In December 2017, WWT started down this path, according to Bechtel, who was hired to lead the initiative. Since then, the company invested $500 million in its Advanced Technology Center, which includes testing developing the expertise needed to deliver AIOps solutions.
Bechtel has seen customer understanding of AIOps' possibilities evolve in that time. Initially, the channel partner spent most of its time educating potential customers.
"A couple of years ago, the room would be filled with Ph.D.s, and no one else really understood how AIOps worked," Bechtel said.
Progress has been made, but it takes time for MSPs to educate many customers.
"We are investing heavily in creating content, such as podcasts and written content, so customers understand how AIOps solutions may benefit them," Windward Consulting Group's McDermott said.
Leading edge companies are slowly starting to turn to partners to introduce AIOps into their management processes. Most of this work is in the testing and prototyping phase with a few large-scale deployments in areas like event correlation. The new solutions help businesses sift through alerts and identify those in need of immediate action.
AIOps has the potential to improve MSP operations in other ways. The third parties run medium and small business networks and can also use the tools in their own data centers to improve service performance and lower costs.
A final use case is using AIOps to deliver new value-added services to their customers. This option is in an embryonic stage because the tools are large, complex and difficult to deploy. Few MSPs have large sums of money to invest so they understand the technology well enough to deploy it for customers.
AIOps presents new hurdles
Partners also need preparation in order to deliver AIOps solutions. Partners must understand AIOps challenges. System monitoring is a complex area, one with a wide range of management chores, including alerts, anomaly detection, event correlation, diagnostics, root cause analysis and security. Typically, MSPs and enterprises already have a solution or tools to perform each management task, and legacy products often do not easily integrate with other systems.
MSPs and enterprises want to break down the management silos and examine their infrastructure at the highest layer possible; however, the integration work is complex and time-consuming.
Sean McDermottPresident and CEO, Windward Consulting Group
"We are seeing progress on AIOps integration because vendors are sitting down at the table, talking with one another and linking their systems," Bechtel said.
Suppliers deliver more comprehensive building blocks, which makes connecting the different components simpler.
Another issue is that staff must learn how to work with the new tools. MSP staff often lack the expertise to collect data, interpret it and train the data model.
Resistance is one more barrier.
"Humans do not like change, and AIOps changes a lot of things: new interfaces, new business processes, new ways of working," McDermott said.
Despite the limitations, channel partners must monitor market developments.
In Gartner's Market Guide for AIOps Platforms, the company stated, "There is no future of IT operations that does not include AIOps. This is due to the rapid growth in data volumes and pace of change (exemplified by rate of application delivery and event-driven business models) that cannot wait on humans to derive insights."
AIOps is in an early stage of development, one that creates many hurdles for channel partners. However, the technology is one that MSPs must monitor because it is gradually becoming a key infrastructure management building block.