The days of managing monolithic style applications running on a single platform are over. Organizations are now committed to delivering their customers a far richer variety of digital services using multiple channels. This means applications are more likely to execute from the cloud, via a multitude of microservices interacting with virtualized resources, containers and software-defined networks.
In this new normal, teams can no longer afford to get bogged down with reactive fire-fighting and lengthy war room sessions. But with so many moving parts, increased application complexity, and dizzying rates of software delivery, what strategies can IT operations employ to prevent being wiped-out by waves of operational big data?
The traditional approach is to buy more monitoring tools; one for every new wave of technology adopted. But this doesn’t scale, negatively impacts margins, and only provides narrow views into the all-important customer experience. So putting tools aside, where should organizations turn?
Well, to the data for starters – or more importantly, to the business problems and opportunities they solve and uncover.
This is hardly an epiphany. Web scale companies understand implicitly the importance of data and gleaning valuable insights. By developing analytics-driven applications, implementing at scale and democratizing usage, these businesses continuously raise the bar in terms of productivity, agility and customer engagement.
In a DevOps context these businesses thrive because their IT teams are equally analytics-driven. Not only do they surpass today’s expectations for delivery speed and quality, they leverage data insights to drive improvements at every stage of the digital service continuum. So before perusing the extensive tools catalog, stop and consider four higher value strategies.
Build an Analytics-driven Culture Within IT
Many teams will collect masses data points; however, what characterizes a strong analytics-driven culture is a focus on collectively leveraging metrics for the benefit of the business as a whole. In a pinch this will involve:
- Empowering IT teams with the applications needed to uncover and share more powerful insights. These can include changes in customer engagement via new mobile app designs, emerging performance/security anomalies, or optimum cloud architecture patterns.
- Incentivizing teams according to business performance goals and outcomes; avoiding persistent “vanity metrics” and operational outputs.
Fast, real-time action or recommendations when insights are uncovered. Nothing demotivates teams faster than finding something valuable and then not being able to act on it.
Democratize Data and Analytics Across the Entire Business
Analytics has limited value when only used by IT operations to support their daily grind. Better methods and techniques treat data as enterprise asset many teams can use, share and leverage in a variety of different contexts. This could involve:
- Delivering an ‘analytics as a service’ operational function where teams can build their own monitoring dashboards and reports to quickly gain the insights they need.
- Ensuring every group is provided with analytical models that have production-level support, together with real-time data that’s ready to use and of known quality.
- Analytics-driven monitoring applications that immediately surface performance insights in-context of different roles and tasks.
Rapid insight prototyping, which when shown to have value quickly becomes established in production processes and tools.
Start Using Analytics Where They’re Most Effective – Customer Experience
You can’t manage what you can’t measure, but collecting and measuring data that’s truly reflective of customer experience is tricky. While some individual metrics will work in specific situations, it’ll be more likely that combos, mashups and new derivations are needed.
In order to gain customer experience insights, analytics-driven applications will deliver teams complete understanding over cloud and on-premise application infrastructure, application performance and the underpinning network. By correlating many metric types across these elements (time series, logs etc.), analytical models become a shared mechanism which DevOps teams use to drive improvements. For example, using predictive models to assess the business outcomes of new code based on application performance or latency improvements.
It’s always possible of course that teams can revert back to a narrow (albeit analytical) perspectives to attack the problem. Some teams will be metric-driven, others will use logs. The trick is understanding how a dovetailed approach will yield greater value. For example, correlating log analytics with network performance management for faster, accurate root-cause determination.
Become an Effective Hunter-Gatherer – Take a Unified, System Level Approach
There’re no disputing many IT organizations have invested time and money in acquiring great set of tools to collect data, so why replace them? What’s really needed, however, is a scalable, open method to aggregate and normalize millions of metrics and logs into one unified data store. Call it an immutable Data Lake, Analytics Warehouse, whatever – having a centralized store of data helps teams quickly search, locate and visualize valuable trends, patterns and correlations. Without this, different groups may resort to managing their own (often overlapping) data sets, using inconsistent access methods, tools and data formats. Fine for a narrow views, but woefully inadequate when teams have to relate data across multiple data stovepipes.
As these strategies illustrate, being analytics-driven is so much more than fixing problems faster. When organizations invest in building an analytics culture, enact new customer-centric methods, and share deep insights, the focus shifts towards treating every problem, pattern and anomaly as an opportunity to improve customer experience.
So stop getting wiped out by waves of data. Start using analytics-driven applications and surf towards a brighter business future.
Opinions expressed in the article above do not necessarily reflect the opinions of Data Center Knowledge and Penton.