AIOps is artificial intelligence for IT operations and yes, it stands to transform IT and operations massively. The deployment of AIOps lets infrastructure owners use a vast array of real time data, algorithmic insight and machine learning to get the very best from optimizing on private and public clouds and automating the way companies migrate applications and workloads to the cloud or next generation platforms. In this article, I will explore how that can be achieved, and how AIOps can be seamlessly integrated into all forms and combinations of data centre(s).
AIOps, the cloud and the data centre
The past few years have seen mass migration to the cloud, and it’s now commonplace for an organisation to use cloud storage and cloud-based applications daily. Much of this data is held and managed in the public cloud (Azure, GCP, AWS, etc) and PaaS and SaaS providers. The public cloud is typically coupled with on premise infrastructure being managed by in house IT teams, third-party providers (System Integrators/MSP’s etc) and colocation hosting providers. Hybrid clouds, which make up this mix of both on-premises and cloud arrangements are increasingly the way enterprise organisations consume IT.
The benefits of moving to the cloud are well understood; business agility, massive scale, ultimate flexibility and organisations not to be consumed with spending huge capex and opex to run their own data centres.
The widespread adoption of the cloud however and resulting hybrid cloud environments, has generated a highly complex IT landscape that now requires infrastructure owners to integrate, monitor and maintain multiple applications, infrastructures and locations simultaneously. Machine learning, and AI algorithms, now permit the automation of many routine tasks, but the management of this technology must also be incorporated into the organisation’s IT domain, and this challenge will grow as the use of AI expands.
As the options for hybrid/public/on-premises data centres have proliferated, organisations have moved between them according to specific business needs. This means that cloud migration is not a one-time event. We read about significant reassessment of cloud consumption due to high costs, performance issues or simply, the migration of a very complex infrastructure to cloud/colo being too much of a heavy lift. As the cost models improve, technology such as containers making migration easier and increased agility the cloud provides, we will see continued migration to and from cloud for the foreseeable future.
For IT teams, multiple clouds and migrations result in a plethora of management paradigms and are extremely difficult to manage, let alone optimise. There are diverse hosts and/or applications being managed through a wide range of tools and accessed through different dashboards. These lack natural synergies and they are not context aware. For example, if there is an outage in one silo and an application performance issue in the cloud the fact the tools cannot speak to each other requires a lot of manual intervention to try and root cause problems – which is a pity, because if those synergies could be realised, providing an overview of the entire IT infrastructure and its functioning from a single viewpoint, the business impact would be huge; reduced outages, increased productivity and the resulting positive revenue impact.
Well, now such synergies can be realised – through AIOps – and the potential is indeed vast.
AIOps brings it all together
The evolution of IT, particularly in the last few years, has resembled that of the motor car. In its early days, the car was managed, driven and maintained by the owner, with input from a third-party specialist. Today, the car is driven by its owner, and still has attention from the specialist, but much of the process is augmented by an on-board computer that keeps it running and diagnoses faults.
AIOps gives infrastructure owners capabilities comparable with those of specialists working on modern cars who rely on tooling and diagnostics to highlight the problem; the machine augments the technician providing insight across all tiers of infrastructure, regardless of location or data centre type, via a single interface.
Using AIOps this way provides two key benefits: The ability to see all applications and functionality in real time and in context and allowing the organisation to pre-empt outages and issues that the AIOps algorithms detect, and to use AIOps real time analytics to optimise choices around operations and infrastructure and what application and workloads should be moved to the cloud, based on algorithmic insight and reliable predictions of future consumption.
The beauty of AIOps lies in its ability to cut through the ‘noise’ generated by the many moving parts of modern IT infrastructure, and show clearly what is working and what is not (or may not, in the future). This gives IT teams the power to predict and avoid outages based on historical data, to expedite and guarantee successful cloud migrations and to make real time decisions around workload and application placement. This, in turn, lets the organisation get the best from cloud capability, maximise data centre value for money and optimise infrastructure resource/capital spend.
Understandably, some are reluctant to throw an entire business behind this new concept straight away – although the signs suggest that AIOps, rather like driverless cars, will in time become the new normal. But a gradual introduction is in any case perfectly feasible — this capability can be applied across the board or used with a few initial applications and then scaled up in ‘baby steps’, according to business objectives.
In short, AIOps is no mere buzzword or ‘next big thing’, but a transformative step in the evolution of IT. And since it will make IT managers’ jobs easier, more efficient and – hopefully – more appreciated, it is surely a step to be welcomed.