Things move quickly in business today. And organizations are increasingly using data to move their businesses forward. IT systems are generating a growing variety, velocity and volume of data. This both creates challenges and opens up new opportunities. Taking advantage of data to reach business goals requires the ability to scale IT operations quickly. AI-led operations (AIOps) helps businesses scale by using artificial intelligence to provide automated, data-driven insights and visibility into operations.
Gartner says, “The use of AIOps platforms to augment IT functions such as event correlation and analysis, anomaly detection, root cause analysis and natural language processing is growing rapidly.” It estimates the size of this market at between $300 million and $500 million annually.
For AIOps efforts to meet their potential, businesses need to define the value they’re seeking. They must put in place the infrastructure, people and processes to ensure AIOps success. And they should implement AIOps in a phased and programmatic approach.
While there are many ways to implement AIOps, including out-of-the-box solutions, an organization’s implementation must match its business needs. A bad implementation is often worse than no implementation at all.
Shifting To A Proactive Stance
AIOps can serve to reduce the time it takes to solve IT issues. It also lowers the noise that IT operational staff faces in addressing tickets and incidents.
More importantly, AIOps shifts IT operations’ posture from reactive to proactive.
AI can provide insights into the data much more quickly than humans can and becomes an even more powerful tool when the insights are connected to automation. With these tools, organizations can free up IT resources from simply keeping the lights on to a position of innovation.
Leading With Clear Objectives
You may want to reach a certain mean time to resolution. Perhaps you want to reduce your ticket expenses by 20% or lower costs by rightsizing capacity. Or maybe you want to free up level-one engineers to do higher-order work.
Feeding AIOps With Data
To put AI-led operations into practice, you need to have operational data — the more, the better. Data comes from many sources: service/incident tickets, help desk tickets, application, server logs as well as metrics and cloud provider logs — all rich sources of information rife for learning.
It also makes sense to bring data from upfront investments like service contracts into the mix. That might include the fact that your organization has procured a contract to buy 20,000 virtual machines. Now you can see that somebody spent a lot of money on the data center. You can then use that financial data as a benchmark against which to model savings. Understanding business goals has to be taken into account at every step of the process.
Gathering The Necessary Talent
Cloud providers are investing billions to make it easier to launch AI and machine learning initiatives. Now, you don’t need as many people, thanks to all the cognitive services and automations of these cloud providers. Just make sure to leverage cloud and cloud-native infrastructure capabilities.
However, you will need both technical talent and subject matter experts. Data scientists develop and test the machine learning models. But data scientists are hard to get. If an organization is looking to fill a data scientist position, it might consider launching an AIOps challenge through Google’s Kaggle community. If you want to launch a Kaggle-type effort, announce your challenge, and let the talent identify themselves as experts through their work product.
Or you can go to a very specialized partner, like a large multisourcing service integration firm. Make sure your partner has a machine learning practice. If they don’t, chances are that they won’t be able to help you. You need practitioners that have experience in this area.
You will also need data operations. The data operations team knows how to manage the life cycle of the data, including processing and preparing the data for widespread use, and is responsible for the supporting tools and the data governance in production.
Data scientists have strong mathematical, statistical backgrounds, but they will not know much about IT infrastructure. Typically, your own IT operations person or team can help with that. Your IT operations team can also provide subject matter expertise and insights into where intelligence and automation can make the most impact. Finally, you need somebody on the team who can automate processes and connect them to your ticket and incident systems.
Letting IT Fly
Now it’s time to get started. Put a pilot in place. Identify some use cases. Develop the models. Run some metrics.
Prove out the business value and KPIs. Test model accuracy, and make adjustments based on the pilot learnings before going to wide-scale production.
The insights you gain during the AIOps pilot and production stages could relate to infrastructure capacity. Or they could provide you with a greater ability to understand the security posture of infrastructure.
For example, you can forecast how and when capacity is likely to run out. Or you can predict where the infrastructure is going to have underutilized capacity. That way, you can shift or rightsize to get more bang for your buck.
AIOps also means you can now go and spot anomalies in your infrastructure. You can even detect potential security issue evidence points. Then you can contain your security blast radius.
Using Predictions To Take Action
It’s very important that all these predictive insights lead to an attended or unattended action.
That means making a recommendation to an engineer or automatically launching a bot to resolve. That way, before something becomes an issue, you can detect it. You can then map it to a certain process and enforce a policy by taking action.
AI Is Here
The current approach to IT operations is not scalable or sustainable. It’s better for your reputation and more affordable for you to resolve issues before they result in outages.
The time for AIOps is now.