Right now one of the hottest buzzwords is AIOps. It’s being used to describe everything from enabling technologies—operational data and analytics platforms—to the tooling needed to automate responses, a.k.a. automation.
Too often the term AIOps is conflated with just one of these technologies, when really AIOps is an overarching approach to operations with the singular focus to enable adaptive applications. That is, applications that adapt to conditions to maintain availability, optimize performance, and ensure security.
Indeed, AIOps is not just about automation. Neither is it just about telemetry (operational data) or insights uncovered by machine learning.
It definitely is not about getting rid of people. AIOps is increasingly important as a force-multiplier to enable a manageable operational staff to handle the growing portfolio of a digital business.
When it comes to operations, AI is about enabling operations to effectively scale, secure, and deliver the growing portfolio of digital services and products necessary for business to operate in a digital-as-default world. Operations needs machine assistance for two reasons:
Even with modern operational approaches (SRE, cloud), the changes introduced by operations are still primarily driven by human beings. Operational decisions are made by humans, codified by humans, and pushed into production by humans. Much in the same way applications developed using agile methodologies eventually collided with transactional (traditional) deployment models to deliver digital services, our modes are mixed. Configuration changes are communicated in a very transactional way, with human beings completing a “form” and then submitting it to an automated system for deployment.
It’s no surprise that 63% of executives say there are too many manual processes involved when setting pipeline infrastructure. Our own research in 2021 found that a disconcertingly low percentage of organizations operate infrastructure and applications today using SRE practices. And yet automating these processes would be preferable to IT and business leaders. The technical capabilities exist today. The ability of systems to ingest and process pertinent data and then formulate an appropriate policy is extant. Combined with the ability to act on those policies through APIs that adjust operating conditions, automation of the remaining manual processes in an otherwise continuous pipeline are well within reach.
However, the ability to adapt automatically requires significant change across all of IT. It cannot be simply bolted on, as we’ve tried to do in the past. For example, to address the need for telemetry we’ve long relied on traditional monitoring solutions that require agents and simulation. This approach adds unacceptable overhead in the form of operational costs and time when workloads are migratory and, in the case of containers, ephemeral. Native instrumentation in infrastructure, platforms, and applications must be the norm to support a real-time adaptive architecture.
Likewise, we have seen the failure of security initiatives and solutions to stop emerging attacks because they are too often stopgap measures that are inserted only after the threat is discovered. Native security and overarching governance are necessary to ensure adaptability of security measures to the constant threat posed by evolving attacks.
Thus, evolution of operations is needed. A mixed operational model will not scale to serve a fully digitized business. The disruption caused by reliance on manual, human-driven decisions and tasks into an automated system makes it impossible to predict time to change and introduces risk from human error.
To address this inefficiency and mitigate the risk, a more adaptive approach to architecture is necessary. One that is destined to see applications grow, shrink, defend, and heal themselves as needed with little to no human intervention.
The defining characteristic of the next evolution of operations is adaptability, and we call it Adaptive Apps.
Adaptive Apps is not a product. It can’t be bought in a box and deployed on a server. It’s an architectural approach to reshaping operations with a focus on enabling the adaptability of digital organization. In other words, the applications and services that are the digital face of business today.
This is no small change. The disruption caused by the first wave of the Internet as organizations raced to “get online” will happen again as organizations push toward operating as a fully digital entity. Operational data must be generated from every system involved (instrumentation) and ingested into a consolidated data platform where machine learning will analyze it to uncover hidden relationships and patterns that result in insights and information critical to the security, availability, and performance of applications and digital services. That information, those insights, must subsequently drive changes into the infrastructure, applications, and systems that adjust—automatically—polices and configurations that stop attacks, address failure, and optimize performance.
These capabilities are expansive and will take more effort than simply migrating to cloud or distributing to the edge. The heart of a digital business—the enterprise architecture—must modernize to incorporate those elements that were not included because they did not exist in the before times, but must exist now to support machine learning, distributed applications, and automation across core, cloud, and edge.
Interested in learning more about the latest ideas and innovations around security, development, and infrastructure for adaptive applications? Register for F5's Agility 2022 virtual event on February 15-16.