When Leicester City Football Club won the English Premier League in 2016 most soccer aficionados were shocked. How could a team on the brink of relegation the year before and without the spending clout of more famous clubs come from nowhere to win the ultimate prize?
We could spend weeks analyzing data about players, tactics, weather, injuries, referee decisions, fouls - heck even crowd noise and fan behavior in the hope of drawing some meaningful conclusions. However, the simple truth is this - Leicester's success was down to having more shots and converting more goals. Sure, data holds a lot of clues but it takes asking the right questions to find the right answers.
Asking the right questions is really tough in IT Operations.
Like football related data, IT teams have masses of information they can process in the hope of gaining meaningful insights. Metrics, logs, topology maps, transactions, events and alarms, all generated from a complex technology fabric and waiting to be questioned by a plethora of monitoring tools. But can these tools surface the answers IT needs to collectively enact improvements that positively impact business?
Probably not – not because they're bad, but because they can often only ask a very limited set of questions.
As we collect operational data, the traditional approach has been to organize in functional buckets. Network data for the network teams, app data for the app support teams and so on. Naturally, tools have matured with new features constantly added to support advances within a particular domain – for example, advanced network traffic analysis, container monitoring – you name it, eventually it'll be covered. We're even seeing many tools incorporate analytics, but even the greatest math can only deliver incomplete answers when processing partial sets of data – a silos is a silo is a silo.
So how did we get here?
The silo'd based approach to data capture and monitoring was fine when technology only needed to support infrequent customer engagements via monolithic apps across a single network. But that's all gone – now digitally empowered customers call the shots, meaning business must continuously strive to craft flawless experiences from many more digital interactions across multiple channels supporting many customer journeys. Operationally this is a tough gig.
More interactions across more channels requires deeper insights and situational awareness that silo'd based data can't surface. Consider for example a situation where a network team detects an alert that's going to result in increased latency for web traffic. With a legacy approach teams can identify that the issue will occur, and perhaps even predict when it will happen. But all these smarts cannot answer more important questions – such as which customer journeys or experiences will be impacted. This can only be answered if the network data is passed in context to another team and combined with more data. It's analogous to a soccer example where the fitness staff determines a goalkeeper is carrying an injury but can't pass the information onto the coach before team selection on cup final day.
IT Operations Analytics ala 2.0, everyone needs to rethink how data should be architected, combined and correlated so that the right questions can be asked of it. This starts by accepting the non-trivial nature of the problem. The inseparability of physical and digital will spawn a new breed of applications generating masses of structured and unstructured data at a scale we can't begin to imagine. This demands forgoing unsustainable data silos with agnostic, extensible platforms that can continuously ingest, store and visualize insights that benefit the entire organization.
Asking the right questions will still require great monitoring tools, only they too will become less domain specific. They'll still support discrete functional use-cases, but will increasingly become capable of accessing and correlating data across apps, infrastructure and networks. So, revisiting the example described above, site reliability engineers could access analytics-driven monitoring dashboards that capture network performance events and present it in context to application topologies and service views. The result is the business impact nirvana pundits have pontificated over for years, but in truth the benefits are much more profound.
So imagine – analytics-driven applications that cross-functional DevOps teams use to identify which application coding practices correlate to the best customer experiences. Or a system that uses predictive techniques to guide development decisions before any platform is prescribed or line of code written. And why not an application that makes real-time investment decisions by continually optimizes cloud workloads based on demand. Only then can monitoring evolve from a break-fix to what it should be - improve-extend – helping every stakeholder across our own digital software factories ask the right questions to get the right answers that lead to powerful insights ...
… and like Leicester City in 2016 start kicking some serious goals.
To learn more about asking the right questions from your operational big data, attend the virtual summit at www.ITOA2Summit.com or download the Monitoring Redefined: Digital Experience Insights white paper.