Back-of-the-Envelope Estimation: Storage Made Simple
In the previous post, we learned how to estimate system load using QPS. Now let’s answer another critical question: How much data will the system store over time? This is where storage estimation comes in.
In the previous post, we learned how to estimate system load using QPS.
Now let’s answer another critical question:
How much data will the system store over time?
This is where storage estimation comes in.
🚀 Why Storage Estimation Matters
Even if your system handles traffic well:
- data keeps growing
- storage costs increase
- databases slow down
So you need to estimate:
how much data you generate per day, month, and year
🧠 The Core Idea
Every storage estimation follows a simple formula:
Total Storage = Number of Users × Data per User
🔢 Basic Example
Let’s say:
- 1 million users
- each generates 10 KB data per day
Then:
1,000,000 × 10 KB = 10,000,000 KB
📦 Convert to Larger Units
To make sense of big numbers:
- 1 MB ≈ 1,000 KB
- 1 GB ≈ 1,000 MB
So:
10,000,000 KB ≈ 10,000 MB ≈ 10 GB/day
📅 Think in Time (Very Important)
Once you have daily storage:
- per month → × 30
- per year → × 365
So:
- 10 GB/day → ~300 GB/month
- → ~3.6 TB/year
⚡ Real Example
Let’s take a more realistic case:
- 100 million users
- each generates 5 KB/day
Using the formula:
100M × 5 KB = 500M KB ≈ 500 GB/day
That’s:
500 GB of new data every single day
⚠️ Common Mistakes
❌ Ignoring time factor
Always convert:
per day → per month → per year
❌ Forgetting unit conversion
Keep it simple:
- KB → MB → GB → TB
❌ Underestimating growth
Data grows continuously—design for scale.
🧠 Key Insight
Storage estimation is not about exact numbers.
It’s about understanding:
“Is this system generating MBs, GBs, or TBs per day?”
🔗 Connecting with QPS
Now you have:
- QPS → how many requests
- Storage → how much data
Next, we combine them to answer:
how much data flows per second
👉 What’s next?
In the next post, we’ll cover:
- Bandwidth estimation
- How data moves through the system
- And how to identify bottlenecks
If this helped, share your thoughts 👇
Originally published by Dev.to WebDev. Aggregated on AIWithGhost for educational purposes — full credit and traffic to the original publisher.