When one works in technology, it’s easy to assume these days that everything in life is built using Docker and Kubernetes.
If you step outside our little tech-bubble world, though, there are actually quite a few examples of organizations that have mastered Continuous Delivery in their own domain. McLaren Automotive is one organization. In fact, it was Britain’s fastest growing company in 2017.
Why has McLaren Automotive been so successful? The answer is simple: everything they do is continuous and measurable.
Each new product has clear objectives, and the design of each product is centered around those objectives. When McLaren release a new car, they already know how it will perform relative to their industry peers. This is no different to the McLaren F1 team that measures and quantifies everything they do.
Continuous Delivery isn’t just about how frequently you can ship stuff; it’s about the impact this activity has on your business so that you’re always improving. More importantly, what is success for every product, release, and deployment?
McLarens First Car (MP4-12C)
If you ignore the McLaren F1 and the Mercedes SLR, the MP4-12C was McLaren’s first crack at being in the automotive business. Its rival was Ferrari, who at the time, had released the 458 Italia.
Ron Dennis (CEO of McLaren Group at the time) said during the launch: “It’s better in every parameter we can measure” than its rivals and peers. Ron was right: the MP4-12C destroyed the 458 under acceleration.
The key point here is about knowing what success looks like from the start of your epics and stories, and continuously measuring what matters throughout your deployment pipeline. You should know in dev, QA, and staging how your application and services behave, and more importantly, how they will impact your business and users in production.
At Harness, we call this Continuous Verification. We let organizations see how their KPIs are being impacted throughout each stage of their deployment pipelines.
McLaren Latest Car (Senna)
The main goal of the Senna was to set the fastest lap times for a road-legal car.
The three objectives of the Senna project was:
- 800PS of power
- 800NM of torque
- 800Kg of downforce
The designers and engineers then went away and created this:
If I asked you right now what the three KPIs are for the current epic or story you’re working on, what would you say? “Errrr” might be a common KPI.
Rarely do we as software engineers understand what success is, or how we measure it. Availability and uptime is not a measure of success — these concepts have no true business context or meaning. We should think more about business transaction performance, throughput, and errors. Even more importantly, we should think about what our users care about.
Measuring the Business Impact of Deployments
At Harness, nearly all our customers have Application Performance Monitoring (APM) tools that instrument the various KPI of their apps and services.
Harness Continuous Delivery has native integration for the likes of AppDynamics, New Relic, and Dynatrace. Harness automates the analysis of these metrics/KPIs using unsupervised machine learning as customer artifacts progress throughout a deployment pipeline.
For example, the below screenshot shows how the performance of several business transactions in an application increased 500% as a result of a new production deployment:
We can also measure and quantify the exact business impact. In this case, Login response time increased from 31ms to 165ms, a 500% increase.
Measure What Matters To Your End User
I’ll leave you with an example of what happens when you measure the wrong things, or set objectives for your releases without thinking about your end users.
Dodge recently launched the Dodge Challenger Demon with the single objective of hitting a 1/4 mile in under 10 seconds. They marketed it as “Most Horsepower,” Fastest 0-60,” “Fastest 0-100,” and so on for a production car.
A few weeks back, TopGear and Joey from Friends reviewed the Demon. For a real person to hit all of the KPIs which Dodge marketed, they needed to a.) Fill up the car with high octane gas and b.) swap out the front wheels with special narrower wheels that Dodge provides for $1 extra. All very fine for when you turn up at a drag strip, but as you’ll see below, not very fine in the real world:
What you just witnessed was a standard McLaren 720S demolishing a standard Dodge Demon.
Despite the Dodge Demon having 808 horsepower, it was humbled by a car with almost 100 horsepower less. Why?
Because the McLaren weighs 700Kg less and has a gearbox that can automate shifting faster than you can blink your eyes. In the real world, these things matter.
It’s a bit like measuring the “five nines” of server availability and declaring victory, only to realize that a chunk of users can’t complete their transactions because their requests are timing out or throwing exceptions.
Focus on Business Outcomes
Continuous Delivery and innovation are key, but so is how you measure success and verify business outcomes.
McLaren may not roll out new cars every day like we do with software deployments in tech, but they measure the impact of everything they do–and they do so at every stage of the product lifecycle. They can scientifically measure and prove every gain against all their market peers. This is why they are leading the sports car and supercar market in 2018.
If you would like to measure and quantify the business impact of every software release, try Harness Continuous Verification.