Originally started as automating the IT operations tasks, AIOps has moved beyond the rudimentary RPA, event consolidation, noise reduction use cases into mainstream use cases such as root causes analysis, service ticket analytics, anomaly detection, demand forecasting, and capacity planning.
During Data Summit Connect Fall 2020, Andy Thurai, emerging tech strategist, AI consultant, thought leader, Field CTO, Forbes Contributor, discussed “AIOps: The Savior for Digital Business Unplanned Outages.”
Videos of presentations from Data Summit Connect Fall 2020, a free series of data management and analytics webinars presented by DBTA and Big Data Quarterly, are available for viewing on the DBTA YouTube channel.
Digital transformation relies on IT operations, Thurai said. Every business is becoming a digital business especially during the COVID-19 pandemic, which hastened the transformation.
“If IT goes down, your business goes down,” Thurai said. “IT is becoming more agile, very cost effective, automated, and serviceable now.”
Current IT landscape is reactive, not proactive because of the transformations that are currently happening, he said. There are 6 challenges that CIOs are facing today: IT budget is spent on operations versus allowing time to innovate, IT operations face limitations, siloed IT operations, remote workforce, the introduction of DevOps, and unplanned incidents can consume up to 90% of the operations.
Why is this happening? Thurai asked. Because disconnected teams are growing in size and using more monitoring tools and siloed, domain-specific teams are cut off from the rest of the organization. The adaptation of new tools and platforms has increased in scale, fragmentation, and complexity of the IT stack, he said.
“This is costing enterprise more than you think,” Thurai said. “An increased amount of bells and whistles creates a chaotic environment. The cost of unplanned IT downtime for enterprises is $40 million per year.”
Organizations can do a few things to future proof IT Ops teams, he explained. IT operations teams will need to realign much like DevOps to create a single source of “observed” truth. Infusing AI into IT operations to reign in the amount of data businesses create and sift through.
AIOps combines AI and IT operations, he said. The platform to do this should be insightful, explainable, automated, transparent, scalable, customizable, and extendable, he suggested.
However, AIOps can be confusing. AIOps is often perceived as a solution of “alert suppressing or incident management,” but research shows that the value for organizations can go beyond these two areas.
The current landscape includes domain centric AIOps, domain agnostic AIOps, or a DIY-verstion of AIOps which is customized to fit the needs of the business.
“The ultimate step-up is to have IT think like the business and they can with AIOps,” Thurai said.
Thurai said companies should start small by unifying analysis across the IT stack, protecting it and streamlining data collection. Organizations should also automate processes to free up IT time across the business.
After Thurai’s presentation, Lewis Carr, senior director, product marketing and management, Actian, explained how to get the most out of Edge Analytics with the right hybrid data management and integration during his session titled, “Understanding Modern Edge Data Management Requirements.”
“We have this huge growth of IoT intelligence, there are so many different applications that are contributing to this boom of edge intelligence,” Carr said.
To manage this incoming edge intelligence wave he recommended machine learning, edge devices, edge gateways, edge on-premise, and the cloud.
The required set of edge data management capabilities includes sensors, intelligent edge devices, or an intelligent edge gateway, or an edge on-premise/cloud.
There are eight requirements for modern edge data management solutions:
- Capable of managing large sets of persistent data
- Run on multiple operating environments from Andriod/iOS to Windows/Linux
- Multiple access methods including CLI, programming, and scripting languages
- Support varied types of data: JSON, BLOB, traditional structured, etc.
- Single platform for Client, Peer-to-Peer, Client-Server, Internet/Intranet architectures
- Stateful and stateless sharing and stand-alone during periods of disconnectivity
- Handle high-speed, multi-channel data collection
- Plug-in-play with advanced analytics, reporting, and visualization tools and platforms
“You should look at something that can work on as many platforms as possible, something that is modular,” Carr said.