Manage What You Can't See: No Magic Required

Andi Mann, chief technology advocate at Splunk, is a big proponent of 'observability' – knowing what's happening with all systems, services and apps to be able to fix and, eventually, prevent problems.

Johanna Ambrosio, Tech Journalist

March 19, 2018

4 Min Read
Andi Mann

Don't rely on third-party services to keep your customers happy. Instead, take action to ensure you can see inside your entire architecture, from third-party apps and services to your own, to prevent and fix as many problems as possible.

Andi Mann, chief technology advocate at Splunk, says this is doable thanks to the notion of "observability" – building APIs and other mechanisms into virtually everything to make sure you can track what's happening. Mann will be speaking about this concept and others at the upcoming Interop ITX conference. That session is titled, AI for DevOps – Using Metrics, KPIs, and Analytics to Drive Continuous Improvement

devops-shutterstock-thin.jpg

"The initial impression with third-party services was that the providers took care of everything for us," Mann explained. "But that's not true; recent cloud-services outages have shown us that the providers say that we need to build our apps and infrastructure to design in for failure" – for instance, by building cloud instances across at least two geographic areas. This means that "you need to know what's going on, to deal with and recover from" even third-party outages.

After all, Mann said, if your business goes down, despite anyone else's role in the matter, "your customer blames you, and you need to be on top of this."

Amazon, Google and other cloud providers do an excellent job of tracking performance and any issues with their individual services, Mann said. But it's difficult for a customer to see all of the services from any given provider at a glance. The idea is to create a data platform to integrate all of these – from all providers -- into one coherent system that tracks your entire infrastructure.

Observability can help. Every technology component, whether a microservice, container or software as a service, emits what Mann calls "digital exhaust," information ranging from logs to other forms of data, that can be fed into analytics systems to create actionable knowledge. He calls it a common visibility platform, and it should be available to all teams throughout an enterprise.

The DevOps role

As artificial intelligence (AI) becomes more enmeshed with DevOps tools and systems, doing this type of analysis will become easier over time, Mann said.

He likens it to an old concept from the manufacturing world: To "see" inside a pipe, insert a flow gauge inside it and then attach a meter on the outside. Similarly, APIs and other tools can be inserted into apps and systems to allow IT to track speeds and feeds, and possibly prevent problems before they start.

Luckily, Mann said, the DevOps movement has already done a great deal to bring the observability notion to reality, and things will only improve over time. "Historically, IT bought into the management space late in the game. They would retrofit apps and systems with management abilities," Mann explained. Now, with developers and operations working together, "They can see what it means to deliver observable apps." He added that they can work together to build in observability from the design stage instead of adding it on at the end.

Another helpful tenet from DevOps is using feedback loops to drive continuous improvement, he said. "Every week, month, day or hour, you make a small change to improve." Access to the observation analytics "enables this type of continuous feedback."

The rise of the business value metric

The observability movement, however, is still in its early days, according to Mann, and there are other things that will be needed to help it along. "As good as we're getting at automating development, testing and QA, we need to elevate the importance of ops, uplifting the system admin role to that of first-class citizen. I see ops people coming to the fore so that they can contribute to a positive customer experience. We need to make customer success a common goal for everyone," he said.

That becomes easier with the adoption of a business value metric. Yes, it is still crucial to measure and manage speeds and feeds, but "we need to see what we're getting in terms of business value – getting more customers to sign up, making more money, achieving the mission/goal of my organization," Mann explained. "This is the near future of DevOps, to see and get visibility across all teams."

In this world, it's even more important than ever to "collect and bring forward one version of the truth," something made easier by a common analytics platform.

About the Author

Johanna Ambrosio

Tech Journalist

Johanna Ambrosio is an award-winning freelance writer specializing in business and technology. She has been a reporter and an editor in the computer industry for over 25 years, covering virtually every technology topic, starting with 'office automation' in the 1980s, as well as management issues including ROI and how to attract and retain talent. Her work has appeared online and in print, in publications including Application Development Trends, Government Computer News, Crain's New York Business, Investor's Business Daily, InformationWEEK, and the Metrowest Daily News. She formerly worked at Computerworld, for which she held various positions, including online director. She holds a B.S. in technical writing from Polytechnic University in Brooklyn, N.Y., now the Tandon School of Engineering of New York University. She lives with her husband in a Boston suburb. Johanna's samples of her work are at https://www.clippings.me/jambrosio.

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like


More Insights