How to succeed in AIOps 


Legacy tools, products, and approaches aren’t well equipped for the sheer volume, variety, and velocity experienced by today’s complex and connected IT environments. Instead, they combine and collate data and package them up into averages – compromising data fidelity. Our current high-volume data environment necessitates team collaboration and critical thinking as a means of influencing the intelligence engine; as it automatically learns patterns, trends, and habits from customers’ proprietary conditions and provides invaluable insights to teams in real-time.

One of the fundamental differentiators for AIOps platforms is their ability to aggregate all formats of data from various sources and then overlap automated analysis on top of it to empower IT teams to be smarter, more efficient, and proactive. A thorough AIOps strategy requires operations teams to widen their purview of both IT and business initiatives, as they offload repetitive break/mundane tasks and concentrate on strategic projects.

Instead of narrowing your AIOps strategy to one particular aspect of the incident response process, New Relic suggests reinforcing the relationships between each stage of the implementation cycle to establish a more powerful solution. It’s not enough to just focus on faster detection, understanding, response, or follow-up; operations teams need a tool that thinks like their best SREs from a systems point of view.

Here are five pillars of AIOps that will enable customers to successfully apply intelligence and realise business value.

Noise Reduction

Modern software environments pose a number of challenges for organisations but one of the most pressing issues is the influx of event volumes teams are being made to sift through on an ongoing basis. Within a week, operations staff are bombarded by hundreds, if not thousands, of alerts.

With robust AIOps capabilities in place, IT operations teams can better correlate events to reduce noise and boost context. Begin by ingesting data from multiple sources and technologies and aggregating a diversity of data types, including events, logs, metrics, and end-user experience monitoring data in a single, consolidated data repository.

Ultimately, successful event suppression is achieved by differentiating those arising within bands of normalcy from those occurring due to true abnormalities that could impact users. This allows IT operations teams to be notified only when a human action is required by their team.

 Ongoing Improvement

A few years back Gartner estimated the average cost of IT downtime at $5,600 per minute. If this were to be applied to today’s downtime overages, modern companies are in need of better approaches to avoid these interruptions completely. Continual improvement is, therefore, a significant intelligent capability, which brings software engineering teams closer to their overall vision of maximising on team knowledge.

AIOps learns patterns on an ongoing basis and applies learned models against inbound alert streams to understand cascading and parallel impacts. It categorises similar alerts into inferences based on the learning models. IT and DevOps teams can then manage these inferences rather than addressing isolated alerts, minimising the “noise” that users need to plough through in everyday operations.  They can then build these inferences to operate continuously and contextually, supporting a constant CI/CD pipeline.


Once existing, manual workflows are implemented into an AIOps solution to automate and scale them, it is essential that teams assess their value, modify and improve them and develop new ones based on the present, or to address, gaps. AIOps promises the ability to execute what heretofore wasn’t practically feasible, and at a scale and speed that makes previously unrealised analytics opportunities feasible. ITOps will transition from a “practitioner” to an “auditor” role. Teams will now have a greater understanding of how systems are processing data and whether the desired business outcomes are being attained.


Identifying anomalies in order to spot problems and understand trends within infrastructure and applications is a critical use case for AIOps. Detection allows tools to both recognise behaviour that is out of the ordinary (for example, a server is responding slower than usual or uncommon network activity is generated by a breach) and react accordingly. AIOps tools can even take automatic action to resolve incidents after they have identified them – they could instantly block a host or close a port in response to a security threat, or spin up additional instances of an application if they determine that the existing ones are insufficient to meet demand.

This is a key component of an AIOps strategy as it permits software engineering teams to both to detect issues as early as possible, before their customers are affected, and allows them to reduce continuous maintenance of detection configuration. Ultimately, it instils confidence in teams that their piece of the production environment is being monitored correctly and in near real-time.


Most organisations are at the early stages of adopting cloud-native technologies; with the failure modes of these new paradigms still looking somewhat nebulous and not widely advertised. In order for software development teams to succeed in our new digital age, it’s more important than ever to gain visibility into the behaviour of applications. Engineering teams need to be able to effectively and efficiently operate modern software systems. In the long term, as systems become increasingly more complex, the only way to seamlessly implement an AIOps strategy is to automate as many tasks as possible for our customers – assist with and augment those that require human involvement – all in a fully transparent and open way to sustain customer trust and retention. This will, in turn, empower software engineering teams to minimise toil and easily audit, control access, and validate their configuration to increase confidence that it is set up correctly.

The competitive advantage of AIOps is truly realised when it delivers collective intelligence – this knowledge will allow organisations to break through traditional silos – fuelling real, efficient, and meaningful collaboration. In this way, AIOps provides invaluable insights that encourage optimised operations and service levels.

Written By Guy Fighel, GM of AIOps & VP of Product Engineering, New Relic