Flow metrics, often used in agile methodology, can teach us how to differentiate "goodput" (work valuable to the customer) from "throughput" (total work done by the system). In this talk I will go over what flow metrics are and what we can learn from it while designing resilient scalable systems.

In a fast paced development team, it is often difficult to identify blockers in real time. This impacts the ability of the team to deliver value to its customers. A system of consistent and meaningful monitoring of key metrics, such as flow metrics, can help address such blind spots ultimately delivering value to customers in a sustainable manner.

Flow metrics comprises the following
1. Velocity: How much work is done by a team per unit time?
2. Time: How much time something takes to get completed?
3. Efficiency: How much time did the team have to wait for something else to be done?
4. Load: How many items remain open in the task list at the same time?

In essence, flow metrics offer a holistic measure of the performance of software development life-cycle. Specifically, it benchmarks the ability of a software development team to deliver on business and technical requests - much like the systems they build which are meant to deliver on incoming requests efficiently.

Inspired from flow metrics, we can apply a similar set of metrics on our system:
1. Velocity or throughput: How many requests per second do we handle?
2. Time: How much time did we take to process a request?
3. Efficiency: Did a downstream dependency be the reason for our outage?
4. Load: If there are too many requests coming in, can we detect it?

In this talk I will focus on two things
- Why flow metrics matters to live applications at scale - this will focus on understanding the foundations of Flow metrics
- How flow metrics can help design better systems - this will focus on designing a system of measuring actionable metrics (for example, too much load can cause a cascading failure across multiple systems that are dependent on one system)