💬
✉️
📸
▶️
Loading…
Loading…
Capacity Planning
1 of 5
The 10-minute estimate that can save you $50,000/month.
4,050
Avg RPS
sustained load
12,150
Peak RPS
3× spike (events)
17
Servers
needed (w/ headroom)
11 PB
3-yr data
with 3× replication
In 2008, Twitter went down under the load of election night traffic. They had 20,000 requests/second — their servers expected 5,000. Nobody had done the math. Capacity planning is doing that math *before* the crash, not after.
Before you build
Choose the right database, server count, and storage tier from day one. Emergency migrations under live load are 10× harder and cause outages.
Before you scale
Know exactly which component breaks next. Pre-provision it — before your users discover it first. Reactive scaling is always too slow.
Before you overspend
A 10-minute estimate prevents $50,000/month in over-provisioned cloud resources. Most teams skip this and pay silently for years.
In every interview
Every staff/senior system design interview requires back-of-envelope calculations. It is the single most testable skill — and the one most candidates skip.
You don't need exact answers. You need answers within 2× of reality — with your assumptions written down so your whole team argues about the right things, not the wrong ones.
Try it in the Library