Expected Scale & Capacity

We begin with looking back, do you remember the lecture on Requirements Engineering we had in the last section?

There we talked about quantitative non-functional requirements that help you to understand the expected scale of a system.

The metrics we got to know were:

• Daily active users - DAU for short
• Peak active users - A multiple that indicates the height of expected peak loads.
• Activities per user, describing how often users interact with the system per day.
• Request Size, the size of each request that is associated with an activity.
• Read/write Ratio - The proportion of read to write requests.
• Replication - a multiple that tells us how often the same data should be duplicated in our databases, for higher reliability and read scalability.

Now it's time to pick-up where we left off and understand how and why we can use these number to estimate the three prominent capacity figures: Request, Bandwidth and Storage.

Highway Analogy 🚙 🚗

Let's look at an analogy to visualize their relationship.

If you were asked to design a highway instead of a software system,

• Throughput would describe the amount of vehicles expected to travel on the highway every day.
• Bandwidth would describe the amount of lanes the highway has. Thereby, it defines how many vehicles can travel on it in parallel. 🛣️ 🛣️
• Latency is a dependent variable that describes the time it would take a vehicle to travel from its origin to its destination over that highway.⏱️

Metrics of Interest