Section 3: The Catalog Fits in RAM - Scale Estimation & Workload Analysis

Executive Summary: This section derives the module's numbers once, aloud. 3 billion devices with 40 apps each give 120 billion installed instances, 4 billion update downloads a day (13 times the installs) and 420,000 update checks a second at peak. Egress is 52 PB a day, 14.4 Tbps at peak, 77% of it updates. The reveal: the catalog is 80 GB of listings and a 1.8 GB update index; the load lives in update checks and update bytes, not browsing.

Section 2 left a page of requirements and one question: which of them is expensive? The interviewer asks it plainly:

"Before you draw anything - how big is this, and where is it big?"

The Catalog Clerk answers "two million apps, so a big product database." The Staff answer is a chain of multiplications.

The Inputs You Agree On

You: "Let me pin the inputs and derive the rest, so if one is wrong you can see which outputs move."

3 billion active devices, 2.5 billion signed-in accounts. 2 million live apps from 1 million developers. 40 user-installed apps per device. 500 million daily storefront users. Peak is 3x average for every rate.

The multiplication the module hangs on: 3 B devices x 40 apps = 120 billion installed instances. Priya's Pixel 7 in Pune is one device; Trailhead Maps is one of her forty.