Art of Codeartofcode
Lesson 6 of 19

2. Back-of-Envelope Estimation

1
The picture to keep in your head
2
2.1 Why estimate at all?
3
2.2 Which units and sizes should you know?
4
2.3 What time shortcuts make the maths easy?
5
2.4 Which latency numbers should every developer know?
6
2.5 How do you turn daily users into requests per second?
7
2.6 Why does the read/write ratio matter?
8
2.7 Why size for the peak, not the average?
9
2.8 How do you estimate storage over the years?
10
2.9 How do you estimate bandwidth?
11
2.10 How much memory does the cache need?
12
2.11 How many servers do you need?
13
2.12 How do you round, and how do you know you're not way off?
14
Memory card and where the analogy breaks
15
Practice: five levels
16
Interview questions
17
Solutions
18
Sources
19
Code lab
2. Back-of-Envelope Estimation

2.5 How do you turn daily users into requests per second?

3 min read+10 XP

From users to numbers

At the wedding: 500 guests × 3 plates each = 1,500 plates. That's for the whole day. The kitchen cares about how many plates per hour, so you divide by the hours of service.

In the system: Three moves: multiply, divide, multiply.

  1. Multiply daily users by actions per user to get actions per day.
  2. Divide by seconds in a day to get the average per second.
  3. Multiply by a peak factor to get the busiest second (2.7).

For PicShare (assumed): 10 million daily active users, each viewing 50 photos a day.

from 10 million users to about 17,400 reads a second at peak.

Figure: from 10 million users to about 17,400 reads a second at peak, in three moves.

  • Views per day = 10 million × 50 = 500 million.
  • Average = 500 million ÷ 86,400 ≈ 5,800 reads a second. (In your head: 500 million ÷ 100,000 = 5,000. Close enough to start with.)
  • Peak at 3 times the average ≈ 17,400 reads a second.

DAU, exactly: the number of distinct users who use the product at least once in a day. Not page views, not sign-ups. A user who opens the app 10 times is still 1 daily active user, but makes 10 times the requests, which is why you multiply by actions per user.

Memory code: "Times, divide, times." Users × actions, ÷ seconds, × peak.