Covid-19 pandemic is a disruptive event like no other and has impacted many critical enterprise business operations worldwide. Global enterprises capitalize on technology to better manage their business. IT operations are the operational layer of an enterprise’s technology landscape. However, enterprises are prone to disruption as demand, supply, and performance flounder. In this article, the experts share their perspective on how enterprises can future-proof IT operations by adopting AIOps to manage the technology landscape intuitively and avoid disruption. Enterprises use digital technology to integrate applications, manage a massive volume of data, and streamline processes. IT operations constitute the heavy lifting by the IT team across the technology landscape. Business enterprises function with clockwork precision and efficiency when IT operations are seamless and have the ability to scale up without interruption.

To put IT operations in perspective, let us understand how a power utility provides 24/7/365 electricity. The utility needs to align demand for power with power supply. The grid needs to optimize the capacity to manage spikes in consumption. The power utility also needs to provide consumers with a choice to switch between conventional and renewable sources of energy. A digital utility uses a smart grid to ‘keep the lights on’ always. The bedrock of the digital utility is operational excellence in IT operations.

The smart grid automates the operational aspects of IT to provide a reliable power supply. A digital utility incorporates artificial intelligence (AI) to provide forensic intelligence for managing IT operations. AIOps leverages machine learning and data science to ingest events and metrics from IT systems across the business, while identifying anomalies, correlating events, and taking preventive action to avoid disruption.

Essentially, AIOps applies AI to IT operations so that enterprises can smartly run IT and scale up the business. AIOps tools identify early symptoms of IT dysfunction, find issues that have occurred in the past to accelerate resolution, and address system bottlenecks across the IT value chain.

AI in IT operations improves predictability by leveraging past knowledge for the future, thereby ensuring reliability in business operations. In real-time monitoring, the availability of dedicated AI computed resources allow IT to become increasingly agile and responsive in near real-time.

AIOps empowers enterprises with autonomics – the ability to be more self-healing, self- monitoring, self-scaling, and self-managing. From monitoring IT systems to application metrics, AIOps helps IT teams evaluate business metrics in real-time and discover the impact of business on IT systems.

During the global lockdown in the aftermath of the COVID-19 pandemic, AIOps pre-empts the disruption of critical IT systems. By capitalizing on machine learning and data interpretation capabilities, AIOps reduces human intervention while undertaking automated troubleshooting / investigations.


Beginning with AIOps – Breaking down silos

AIOps can streamline and enhance company-wide IT functions, reduce repetitive labor and create new and better processes through data-led insight. While a lot of AIOps products in the market talk about applying AI to a few specific IT cases, AI can be applied in any use case where past data is available. Some of the roles, as defined by ITIL, that could easily and effectively reap the benefits of AIOps include service operations teams like incident management, L1 and L2 support teams, request fulfillment teams, access managers, IT operations control teams, and application management teams. Continual services improvement teams such as process owners and process architects, can do a lot with AIOps

Service design teams like catalogue management, service level management, risk management, capacity and availability management, and compliance team members, can also find great advantages with AIOps.

To elaborate on this further, we’ve listed below seven characteristics that define a true AIOps solution, as mentioned by analysts at Ovum (Ondia). These were featured in a recent eBook1 created by EdgeVerve in collaboration with AIBusiness.

AIOps must be platform-agnostic and operate in all environments, noting that while operating across 100 percent of business environments is “unlikely,” AIOps solutions should be capable of working across any on-premise or cloud infrastructure in terms of x86-based workloads.
To drive collaboration across the IT organization, AIOps must at its heart be easy to use and enable intuitive sharing of information. AIOps should be largely invisible to the people using it, while enhancing existing tools and processes.
AIOps needs data access and an insight-generating element. The AIOps solution is made up of lots of disparate data sources, and it must be able to evaluate the quality of this data and, if needed, to store aggregated versions. It considers a solution’s ability to correlate data and identify new insights as a key benefit of using AIOps that enables IT departments to become faster in identifying issues and resolving problems, leading to their becoming proactive in terms of problem resolution.
Security and compliance management are popular use cases for AIOps. Understanding when equipment is out of compliance or when an unusual event has occurred are key capabilities that can shorten the time from a known incident to a resolution. Understanding the known good behavior of a system is critical to identifying an anomaly, which could be a security breach that needs to be investigated.
In terms of compliance, AI should support an IT department to know the status of the system, workload, and when it should be patched.
The baseline for automation with AIOps refers to the ability to automate many tasks is one way that IT operations can begin to regain control of a complex and fast-moving environment. AIOps should be able to identify tasks that need to be automated. Besides, AIOps should drive automation of more and more tasks from simple sysadmin alerts to process automation, driven by learned behavior. AIOps should ensure that the complexity of automation matches the technological maturity of the wider organization.
According to Ovum, as AIOps becomes mainstream, reporting and analysis expectations will change. Metric-based reporting and analysis will become a part of the process. Ovum illustrates it with a use case: In metric-based reporting, the organization identifies business outcomes and associates relevant metrics. For instance, if a business change to a customer-facing application has a potential value of US$ 100,000 per day in increased revenue, a metric that measures time from concept to production can be linked to this value.
As a unifying technology, AIOps must be “extensible.” According to Ovum, “AIOps is not a suite of solutions that will rip and replace existing management tooling; rather it is a thin management layer that connects all these activities and uses AI technology to improve the business of IT delivery.” AIOps will cover specific areas of complexity and extend to any area of business function in which AI and data-led insight can enhance technology operations.

AIOps: The answer for too much information

One of the biggest challenges facing the IT organization is information overload. An average IT team in a large business enterprise received nearly 3,000 daily alerts and notifications in 2019, according to software intelligence firm Dynatrace.

Making sense of data and prioritizing critical issues is a difficult task. The IT industry realized that sifting through logs and metrics can be outsourced to artificial intelligence models. AIOps was born to address the need to manage and capitalize on big data.

In a recent webinar conducted by EdgeVerve and AI Business, Jasdeep Singh Kaler, Global Product Head of Infosys Nia at EdgeVerve, proposed how to streamline enterprise AI deployments.“The volume of all tickets and metrics and logs is increasing at an extremely high pace. It is humanly impossible to link them together,” Kaler said.

AI models ingest the data produced by hardware and software, discover the critical issues, and recommend the best course of action. AIOps software will become increasingly sophisticated to a point where remediation takes place automatically.

AIOps is the next phase in the evolution of ITOps. Some key differences between the two: Traditionally, ITOps involved automation for solving problems faster and more efficiently. But given the scale, data velocity, and variety, it is very difficult for a rules-based approach to automation.

With AIOps, we are not just talking about solving problems quicker, we are preventing problems using key technologies of AI. AIOps also serves as the integration layer, bringing information from disparate metal boxes together, without having to rely on proprietary vendor tools.

AIOps helps enterprises manage more complexity, more diversity of their application stacks, more telemetry, with fewer resources and more intelligence.

What can enterprises expect from an AIOps solution?

In the short term, IT organizations can look forward to enhanced IT productivity, reduction in the mean time to repair (MTTR), reduction in duplicate tickets, preventing outages, optimizing team effort which is spent, (delivering) other cost savings. In the long term, IT organizations can drive more customer-oriented applications, enhanced business agility, and introduction of new services.

AIOps success story: Infosys Nia AIOps automates payments for a bank

AIOps success story: A bank wanted to improve operational efficiencies in IT services. The enterprise had twin challenges: a tremendous amount of effort was required for highly knowledgeable L2 technical support for Payments service, and several L2 subject matter experts (SMEs) were nearing retirement.

The bank consolidated its Payments ITOps services from multiple vendors into EdgeVerve, leading to swift demand reduction and a simpler daily operating model by applying AI to IT operations.

Infosys Nia AI platform used problem management analytics to identify high impact automation areas. The Infosys team developed nine deterministic cognitive use cases and incorporated automation with an AI-based knowledge model for knowledge management. The skills and expertise of the retiring SMEs were captured in a knowledge base for the future.

AIOps success story: Infosys Nia AIOps transforms service delivery at a telecom company

A telecom company wanted to implement a product order activation capability for complex product bundles such as the Internet, TV services, and phone line subscriptions. The telco sought a bundled offering with zero latency, end-to-end order visibility, order monitoring, automated error and notification management, using AI and machine learning models.

The Infosys team adopted the Nia platform to drive an AI and automation strategy. Events and logs across existing systems were amalgamated for dynamic discovery, and process and order execution. The solution offered a real-time, ‘milestone’-based view of an order process.

An AI-led order activation service produced a unified dashboard view supported by a prediction model. It predicts the likelihood of order fallouts, automatically resolving them as they occur, while providing visibility of customers.

An enterprise needs to manage the sprawl of data and applications as the buiness scales up. The IT team can manage IT operations smartly by adopting advanced automation and data science. AIOps is a catalyst for business growth through smart application of insights from past performance to manage daily IT operations, mitigate unforeseen risks, provide immunity in crises, and plan for future events.


The Human Factor

Impact of AI and Automation on how we live, work, and play

Like what you are reading ?