In the digital transformation era, application development takes a boundary-less approach. Applications are either built in the organization’s own datacenter or leverages IaaS, PaaS, FaaS or public cloud provider services.

As the IT landscape evolves, users expect the applications to be available at all times, across devices. For enterprises, the challenge in delivering to this expectation revolves around the following:

• How can we manage an application that is deployed across multiple environments and still ensure a seamless user experience?
• How do we pre-empt an issue before it affects the end user?
• How do we pinpoint whether the issue is in user’s device, application server, in the middleware or database, or the network that is being used to connect? Or worse, how can we tell if it is a piece of bad code that lies at the heart of the issue?

Often, these challenges stem as a result of support teams running operations with traditional monolithic monitoring tools, updating configuration management database (CMDB) manually, and using siloed approaches to datacenter, network, and application management. Support teams need dynamic and intelligent monitoring and management systems to deal with the current requirements.

This paper introduces you to the building blocks that an Operations Intelligence Platform needs to have in order to run AI-driven operations that effectively support a heterogeneous, multi-cloud, multi-geo platforms where applications are deployed.

AIOps: The next-frontier in IT operations

Today, application components are spontaneously created and destroyed to support dynamic load requirements, or to apply bug fixes/upgrades without impeding user access. At the same time, users access applications on a variety of devices such as mobile, smart watches, smart devices and so on. The monitoring and management components that support this fluid environment must be dynamic enough to understand the changes and provide real-time updates. This is critical to enable support teams to address issues proactively.

To enable such intelligent operations, organizations must re-think the way they conceptualize their requirement for support teams and empower them with a digital cockpit to manage the environment. Artificial Intelligence for IT operations (AIOps)  will be key to managing multi-cloud and multi-vendor heterogeneous environments.

To execute AIOps, a comprehensive analytics and machine learning strategy that leverages holistic data is needed. Automation can then use the resulting intelligent insights to enable continuous improvements. Figure 1 illustrates key building blocks of AIOps.

Key components of AIOps

• Visibility – For the support team to function effectively,  end-to-end visibility into the infrastructure of the application is critical. A top-down configuration-driven discovery solution can help discover the end-to-end application architecture - from the network load balancer where the application access begins to the last mile infrastructure. Once the application architecture is mapped, support teams can migrate, support or optimize it.
• Monitoring –Traditional monolithic, threshold monitoring is no longer adequate in the post-digitization era where customers don’t wait to provide negative feedback. They simply delete an app if they don’t like it. Monitoring must therefore help support teams proactively drive positive end user experience.
• When support teams have visibility into the application architecture, it’s easy for them to ensure holistic monitoring. They can also identify what aspects of monitoring are to be configured from an end-user device for response time monitoring, and what parameters must be monitored at network, application, platform, middleware, and database levels. It’s also important to automate monitoring and absolute thresholding to ensure that service assurance and alerting are in place.
• Event management – Continuous Innovation and Continuous Deployment (CI/CD) are not sustainable without Continuous Assurance (CA). AI-driven event management provides CA by industrializing the ingestion of event data, noise filtering event data, and transforming huge volumes of disparate events into proactively actionable work items - in the form of clusters of alerts. As a result, AI-driven event management increases developer productivity, reduces the time to detect incidents, prevents adverse business impact, and reduces the number of actionable work items along with the mean time to resolution (MTTR).
• Analytics - Advanced analytics applied by collecting unstructured data from various sources helps support teams spot the unknowns for pre-empting certain issues. According to Gartner 80% of issues in an environment occur because of a change being made[i]. A change configuration analytics solution is included in the reference architecture to help pre-empt issues. Change configuration analytics helps identify who made the change and where, and assess the impact of the change.
• Automation – Automation must be applied consistently across operations in an environment to enhance outcomes. The process should begin with discovery and monitoring, and then progress to event management, task and process automation and so on.  AIOps brings 3A’s i.e. AI, analytics and automation together to deliver measurable outcomes.
• Intent-based tooling –Different workload management solutions are critical to managing a heterogeneous environment. Managing the lifecycle of these solutions requires varied skillsets in order to ensure they are delivering the intended results. This can lead to higher overheads. The intent-based tooling system acts as a “monitor to the monitors” to manage the different workload management tools in the environment. It enables automated discovery of agents and provides a holistic view that helps administrators define central agent management policies, auto-fix and reconfigure agents, and orchestrate the agents and servers they report to. Diagnostics around agent performance provides critical insights for effectiveagent resource management and throttling.

With this centralized control in place, administrators can pass on the intent (configuration) to the tools  for execution. This eases the workload of administrators to a large extent. Above all, it helps maintain uniformity and centralize management.

Making collaborative IT operations a reality

AIOps  helps organizations move from an egosystem of silo-ed apps, datacenters, network, and IoT teams to an ecosystem where collaboration reigns. An ecosystem-driven IT brings various teams together to deliver incremental value to enterprises as well as end users.

Reference

i  https://www.gartner.com/doc/2939717/top-strategies-optimize-change-management

Guharam Sugavanam

Practice Development Head – AIOps & BSM Tools, Wipro

Guharam is a senior IT professional having more than 16 years of industry experience across the technology spectrum - presales to implementation, strategy to hands-on, hardware to application support. His current focus area is to enable innovative solutions on AIOps & BSM tools. He had led cross-functional and cross-geographical teams on large transformation engagements on Service Management, Service Assurance, Automation, AI and Analytics.