When I first started learning system design, one thing kept confusing me:
How do people quickly estimate system scale without exact data?
You’ll often see statements like:
- “This system handles ~10K requests per second”
- “We’ll need ~500 GB per day”
- “Bandwidth could be the bottleneck”
But where do these numbers come from?
That’s where back-of-the-envelope estimation comes in.
🚀 What is Back-of-the-Envelope Estimation?
It’s a way to quickly approximate system scale using simple assumptions, without needing exact data.
The goal is not precision.
The goal is:
Get close enough (within 10x range) to make design decisions.
🤔 Why does this matter?
In real-world systems:
- You rarely have exact numbers
- You still need to design scalable architecture
- Over-engineering is expensive
- Under-estimating causes failures
So instead of guessing, engineers:
Break problems into small assumptions and estimate logically
🧠 The Core Idea
Every estimation follows a simple chain:
Users → Requests → Data → System Load
You start with something you can assume, and build from there.
⚙️ The 3 Things You Always Estimate
Almost every system design problem boils down to:
1. Traffic (Requests per second)
How many requests is the system handling?
2. Storage
How much data is being stored daily/yearly?
3. Bandwidth
How much data is flowing per second?
⏱️ The Most Important Approximation
To convert daily traffic into real-time load:
- 1 day ≈ 100,000 seconds
This helps you quickly estimate requests per second.
🔑 Example (Simple)
If a system gets:
- 10 million requests per day
Then:
10,000,000 / 100,000 ≈ 100 requests per second
That’s your working estimate.
⚠️ Important Mindset Shift
Back-of-the-envelope estimation is NOT about:
- exact calculations
- perfect accuracy
It IS about:
- structured thinking
- reasonable assumptions
- fast approximation
🧩 Why beginners struggle
Most beginners:
- try to be too accurate
- don’t know what to assume
- get stuck on numbers
Instead, you should:
focus on structure first, numbers later
🧠 Key takeaway
You’re not trying to calculate perfectly.
You’re trying to answer:
“Is this system small, medium, or massive?”
👉 What’s next?
In Part 2, we’ll go deeper into:
- Requests per second (QPS)
- Mental shortcuts for fast estimation
- Real examples
If you found this helpful, feel free to share your thoughts 👇
Top comments (0)