Understanding Full Stack Observability

, Oct 26, 2021

What is it and how do we get started?

Whilst Observability is a current buzzword, I was reminded of Control Theory lectures many years ago.  At that stage it referred to knowing the internal state of a system from observing the outcome.  It relied on the systems being linear and deterministic.  But our connected world is anything but linear or deterministic.  It is hugely interconnected in complex meshes, processes like those involved in online purchasing, run independently but are interdependent, and AI will result in non-deterministic outcomes.  My old understanding of the term does not work in this new world.

So, Observability has been re-invented, re-defined.  It has also been extended with concepts like “Business Observability” and “Full Stack Observability” emerging.  Why?

It is all down to your interest – it defines the viewpoint and outcomes you want to observe and influence or manage.  Examples include:

  • The CIO wanting to understand revenue at risk and losses during an incident
  • The merchandising team wanting to understand customers application experience online and in store
  • The Devops team wanting to understand that the new release does not degrade user experience
  • The infrastructure manager optimising the mix of cloud and data centre hosting
  • The head of cyber security understanding how to manage a threat
  • The service desk understanding user experience for remote staff as they travel

Remembering the adage – if your only tool is a hammer, every problem requires a nail – its clear that with such a wide range of outcomes we will need a whole bag of observability tools.  And, as the number of complexities and viewpoints grow, so will the demands on the toolset.

Today we see Full-Stack Observability as supporting the seamless digital experience of technology by providing Observability across the full stack of application, infrastructure, network, and security with real-time insights correlated across domains, with business context, all powered by Artificial Intelligence / Machine Learning (AI/ML).

Trying to understand the outcomes required rather than the lens you observe through, the outcomes fall into 3 categories:

  • Performance – a view of the current (and historic) performance of every aspect of the service – even those beyond your control to support operational management.
  • Optimisation – of both the service delivery and cost in order to ensure both meet the needs of the business.
  • Security – using security metrics and service metrics to detect and manage threats and anomalous behaviour.

To extract the data from the system to support the different viewpoints and the three sets of outcomes requires a toolset.  Our Partners at Cisco are working on toolsets and the Full Stack Observability approach complements their SecureX initiative to bring all security data together in one dashboard.  From a Cisco standpoint Full Stack is currently:

  • AppDynamics – classically an application performance management tool, now capable of monitoring the end user experience, looking holistically at the application, database and server and delivering a business view, including revenue impact.
  • Intersight Workload Optimizer (IWO) – provides a deep insight into the application and infrastructure stack, recommending optimisation from a performance and cost perspective across public and private cloud.
  • ThousandEyes – gives you an outside in and inside out view of the end user experience from a network perspective including the complexity of the Internet and your own networks.
  • Integration between the components – some “out of the box” API’s are available, but extending the stack through iterative integration sits with the integrator.

But that looks at a very traditional “Full Stack” which is already developing with containers and micro-services.  No doubt mindful of this Cisco’s acquisition, closed in October, of Epsagon – focused on AWS – to monitors cloud, container and micro-services and the recent announcement of its intention to acquire replex to give Kubernetes insight.  As these integrate and expand technology coverage it will keep FSO relevant.  Supporting the ongoing innovation around integration across the toolset will be a value multiplier.

A business can now gain insight into the systems that define its’ offering, from the business outcome, through the “full stack” – the application down to the network layer, and from the point of use through to the application and database.  Harvesting this data constantly and using AI and visualisation to make sense of it is the current state, but before long the end user experience will be optimised automatically using this telemetry.  This is being accelerated through integration of a number of leading products to give customers the early benefit of Full Stack Observability – a better end user experience.

To find out more about how full stack observability can benefit your business, please reach out to a member of the team.

Related Insights