In all aspects of life a small advantage can make a huge difference. Consider US high jumper Dick Fosbury at the 1968 Olympic Games. With a unique technique where he leapt backwards over the bar, Fosbury's method gave him a much lower center of mass than the traditional straddle techniques. Fosbury snagged the gold, with good old physics helping him do it. Today, high-jumpers routinely use what's now known as the "Fosbury Flop" — a great analogy for digital business, where new best practices are adopted rapidly and competitive edges tough to maintain.
Like Fosbury, businesses in every vertical are in race to gain a competitive edge, understanding that anything in a digital world has a limited shelf-life. To do this they're turning to big data and analytics; gleaning valuable information and then pivoting quickly before the pack catches up — essentially they win because their insights are actionable.
So, for example, a pharmaceutical company might use an analytics platform to predict the spread of seasonal influenza, ensuring that clinicians in the effected regions have needed vaccine. Or, a large sports equipment retailer might automagically sift through masses of weather and social data to determine where the first winter snow will fall to ensure stock is on hand when marketing campaigns are run. In both cases: insights leading to actions.
The IT Operations Analytics Bar – How High Can You Jump?
In IT Operations, we're also on the hunt for data-driven insights and they're not a million miles away (nor should they be) from those of the business. With success hinging on the all-important customer experience, any anomalies that could impact digital engagement must be systematically identified before any business advantage is lost.
Similarly, as seasonal customer demand increases through innovative analytics led marketing campaigns, operations needs ways to accurately predict the optimum cloud compute resources needed to maintain performance. In such situations "best guess" manual methods or partial insights have dire consequences. Predict too low and teams will be scrambling to put out performance spot-fires, too high, and profit margins fall because of unnecessary infrastructure scaling.
It sounds easy paper, but gaining these types of insights is a tough operational gig. A deluge of data streams in various shapes and sizes (time-series data, quality metrics and unstructured logs) impedes visibility. And, with teams organized around technology stovepipes, it's no surprise that attempting to address the problem through a narrow diagnostic lens has become the norm.
Keyhole Views into Data Limit Insights
It's not through lack of tools that organizations struggle. Many IT ops team routinely employ a variety of products to gain visibility, with many providing analytical capabilities. However, as technically good these solutions are, they only provide keyhole-views into problems and lack the much needed business context to reveal deeper insights.
But perhaps the bigger problem is that any hope of gaining advantage from real-time insights (even in a narrow purview) is lost because teams lack the ability to act on them. Even if an anomaly is distilled from mountains of events, alarms and time-series data points, any actions (be that for remediation or optimization purposes) must be executed faster than teams can respond. It's critical therefore that making full use of insights requires monitoring solutions and processes that not only surface the insights themselves but act on them immediately. No clearer is this more evident than in the area of application performance management (APM).
Modern APM solutions should employ the right math and machine learning to accurately pinpoint anomalies that are "true-positives" because these are the only event that is truly actionable (false-positives lead to staff burnout; false-negatives business burnout). But actionable insight is more than just accurate anomaly detection. It also means ranking and prioritizing by business impact, together with methods to correlate application to infrastructure to speed root cause. Combine this with automated ways to deliver assisted triage within the whole process and insights become very actionable.
Make Data Work for IT Team and Not the Reverse
Of course gaining insights consistently also means adopting solutions that are flexible. In reality, all anomalies are contextual and what might be considered bad in a customer-facing web or mobile scenario might not be for an internal business app. It's why modern solutions employ methods that can marry the context to the math in order to increase productivity and prevent an explosion in one-off costly analytics chasing.
Actionable insights from analytics aren't just tactical, they also provide operations professionals the intelligence needed to shape positive decision making across the organization. Imagine for example, sharing predictive performance modeling capabilities with developers to help drive better designs and code quality, or working with finance to optimize cost structures by predicting and matching cloud compute instances to workloads.
Whether he knew it or not, Dick Fosbury used physics and math to gain some important athletic insights that were immediately actionable. Agile Operations teams work the same way — making data analytics work for the organization as a whole and never resorting to traditional thinking, practices and tools.
Those that do this consistently always raise the bar, gaining new levels of operational intelligence needed to fuel the digital business. And those that don't? Well, they risk another type of "flop" altogether.