2020 Predictions – Centralizing AIOps Platforms
During 2020, deployment of AIOps functionality will migrate to an approach that delivers all algorithmic functions from a logically centralized platform. AIOps can be defined as the sequenced application of algorithms for data selection, pattern discovery, inferencing, communications, and robotics – applied against varied IT Operations use cases. Up to now, solution delivery has largely been piecemeal and data domain specific, an extension of existing management technologies and disciplines.
So why am I predicting centralization?
One of the main issues with AIOps is that you have signals coming through a massively noisy dataset and those signals need to be understood and delivered to the right agents who accordingly respond. Essentially, it’s a question of getting the right signals to the right responders.
If you look at all the various management technologies that have evolved over the last 40 to 50 years, they have either been involved in ingesting signals from multiple different sources or have been associated with receiving the output of these discovery or are ingestion technologies which provide a degree of automation to respond. The response could be human routing to the right level of expertise or it could be a further smart signalling system which bypasses the service desk or it could be some kind of automation capability. The point is you have many tools out there.
AIOps is a bridge
What is needed is a technology that will integrate all of these signals coming in from different sources and make sense of them all, to ensure that these signals get to the right community of agents, both automated and human. But it also a question of assembling and coordinating that community so that it is in a position to respond to the signals. Basically, you need some kind of bridge technology which is moving the completed signal to an agent which is ready to receive it. That bridge in the middle is precisely what AIOps is supposed to do – coordinate on both ends and make sure the signal is received by the appropriate agent.
If you build AIOps on top of either of these different input points or around these different response agents, you can probably improve those individual points but you still have the problem of coordinating the other inputs and coordinating all of the agents responding to those inputs.
What I’m seeing with many enterprises is a refurbishment of many input mechanisms and the rethinking of their response technology, which means moving away from older generation solutions. New technologies are being introduced on the signal reception side and the signal response side, but there is a big blank in the middle which is precisely where AIOps fits. Therefore, the delivery of AIOps as a logically centralized architecture will come to be, even more so as enterprises transition from ITIL3 to ITIL4.
ITIL 4 and AIOps
ITIL 4 and AIOps are fundamentally compatible. In fact, one could go further and argue that the deployment of AIOps technology is a prerequisite for realizing the objectives of ITIL 4.
In previous versions, ITIL had always paid lip service to the need to breakdown technological silos. In point of fact, process boundaries were guarded with severity. This focus on segregation among processes, coupled with a generally infrastructure-centric perspective on IT Operations, ended up reinforcing the existence of silos.
In a dramatic reversal, ITIL V4 acknowledges that, in order to manage modern IT systems in the delivery of value through services, silos must go. It stresses that the IT Operations practitioner must observe, analyse, and be able to modify systems end-to-end — without regard to technological silos or process boundaries across the entire organization.
SEE ALSO: 4 steps to implement a CI/CD pipeline
As a collection of technologies and practices, AIOps helps achieve the goals of supporting continuous service delivery while optimizing business agility. Having all five distinct types of AIOps algorithms acting in concert is the only way to deliver a holistic view of modern IT systems. Their modular, distributed, dynamic, and ephemeral nature makes it impossible for human operators to monitor their performance alone.
Intelligent automation via AIOps software and machines helps to select the data, discover the patterns, make the inferences, communicate the results, and take the required actions. Critical IT incidents are detected, investigated, and resolved quickly. AIOps has become a necessity for delivering this vision, and therefore, a necessary prerequisite for any IT organization taking on ITIL 4.