Data is generated in copious quantities by every device, system, and application powering the digital economy. Whether it's GE collecting 5,000 data points per second from jet engines to transactional data behind the nearly $1 million per minute spent online, data is about to overtake the app and claim ascendancy over the digital economy.
But there's a problem. Data is inherently meaningless. Seriously, it is. Consider this data: red, round, small.
What is it?
It could be a lot of things. A marble. A cherry. A cranberry. A Lego piece.
Data, you see, has no real meaning until it's processed, structured, and enriched by context. Ironically, context is provided by other data. But even if I give you another piece of data, perhaps "food," you still can't be sure what it is. It turns out, you need a lot more data to determine what is being described correctly.
In order to make sense - and strategic use - of data, you need to collect a lot of it from multiple places. In the realm of application delivery, those places are an increasingly distributed set of application services that span the data path from code to customer.
Each of those services provides the perfect place from which to gather operational and application layer data. Now, most application services already emit a lot of data in the form of logs and alerts, but that data is largely ignored or only temporarily stored. The data is important for day to day operations and post-mortem investigations into security or performance incidents, but beyond that is often discarded as non-essential.
But the true value of data can only be derived through the discovery of relationships and patterns across time, whether measured in minutes or months. In all cases, the data points must be extracted and delivered to an analytical platform. There the data can be mined and married to business-oriented data points to produce actionable insights with real technical and business value. Because it isn’t enough to know that a mobile app is slow. It’s important to know why and what the business impacts are. It’s not enough to know that conversion rates are decreasing, or abandonment rates are increasing. Being able to relate poor mobile app performance to those data points provides insights as to why - and how to address them. That relationship can only be discovered through analysis of business and operational data points.
The trick is extracting data points. It's not enough to gather data points from just the web server. It's not even enough to gather data points from the web server and the client. It's necessary to gather data from as many points in the data path as possible, from each of the application services that comprise that path. Without a holistic set of operational data points, finding the cause of a problem will be infinitely more difficult. Consistent, correlated data points across the entire data path are the secret to unlocking the visibility needed for fast discovery and remediation of real business problems.
The best way to accomplish this seemingly monumental task is to instrument as many application services as possible. From the web server to the API gateway, from the service mesh to the ingress controller, from the app access service to auth proxies, and from DNS to the client code, all need to be enabled with the ability to emit telemetry that can be enriched, structured, and correlated across transactions and time. Only then will machine learning and analytics be able to produce the kind of information and insights business and operations needs to extract full value from data.
Today, most application services can't emit the telemetry needed to enable this capability. But in the coming year, you should expect to see a growing number of application services across every operational domain imbued with the ability to emit the telemetry necessary to unlock the business value of data.