By Tania Roblot | SECTION6 Observability Field Operative
Being customer-focused is no longer optional. Every software platform and application critical to business success needs to be driven by the customer and user experience through design, implementation, and support. Unfortunately, understanding the customer and user experience is usually elusive outside of monitoring social media feeds, customer support requests, or the noise generated from a significant outage.
Murphy’s Law dictates that things break. There is no way to predict every scenario and meaningfully catch every possible blip. What does this mean for you? Simply, that monitoring is not sufficient to meet your needs.
Monitoring is designed to catch known behaviours of unknown root causes — or causes beyond our control, such as a third-party dependency. These are often called “known unknowns”, and we catch them by making assumptions about the behaviour and then record this activity in aggregated metrics.
However, when something unexpected happens and your service breaks, your aggregated metrics will not help you identify a root cause. This is because the raw data of what is happening in your system has gone — there is no way to drill down into metrics to reconstruct that data.
So, what do we do if testing and monitoring are lacking to solve this problem?
Enter observability. Observability often gets confused with monitoring. They are “sister” practices that work closely together but fundamentally serve two different purposes and cover different problems.
Unlike monitoring, observability helps look into what are often called “unknown unknowns” — system behaviours that are unexpected and without a known root cause. In a way, observability aims at refitting your black box services with clear panes of glass, so you can finally peer through and watch the cogs move.
Observability aims to empower you to answer any question about what is happening in your services without interacting with them. Having an observability practice in place will allow you to turn unknown unknowns into known knowns — fully identifying and resolving an issue — or known unknowns — growing your monitoring coverage. However, observability isn’t only useful for incident response.
With the ability to ask any question regarding the state of your services, you can track
behaviours in real-time. Meaning your engineers can watch a deployment go live and how it impacts the service. They can immediately see if their change impacts users and roll back that change before someone gets paged — i.e. before it becomes an outage.
Imagine how that might positively impact your deployment frequency, lead time for changes, mean time to repair (MTTR), and change failure rate! Not only that, but with this new kind of insight into your services, you can use them for your feature and development decisions. With an accurate view of your customer experience, you can address performance issues faster than ever before.
Observability will lead you into a world guided by evidence — helping you make better and better decisions towards building a great service.
If this is the kind of story you would like to invest in for your business, then we at SECTION6 would love to help you on this journey.