Wikipedia's system design
☰
Wikipedia is a read-mostly machine: over a thousand page views for every edit, so the whole design is about answering reads from a cache close to you — and throwing exactly the right copies away when someone edits.
+ The whole system in one paragraph− Hide
Wikipedia gets over a thousand page views for every edit, so the design is about answering reads from a cache near you. Cache servers in seven data centers answer most logged-out reads from memory or disk. An edit is written once in a core data center, and a purge message tells every cache server to throw away its old copy.
In this lesson
3 chapters · ~8 min · a quick check in eachEverything is on this page — just scroll down ↓
The whole system
What it must do — and how well
Wikipedia, end to end
Left: the edge (DNS, a load balancer and two caches, in every data center), which answers most reads. Right: the core, where MediaWiki builds pages and databases keep every revision.
You type en.wikipedia.org. Wikipedia's own DNS sees where you are and answers with the closest of its seven data centers.
Read-mostly, in numbers
English Wikipedia got ~1,400 page views per edit in 2025. That ratio is why caching whole pages pays: in 2020 the caches answered most of what they received.
Curious? Go deeper+ 1 optional
Seven data centers, two cores
Five data centers only cache, close to readers; the two cores, Virginia and Texas, run MediaWiki and the databases. An edit goes to the primary core, which swaps twice a year.
A reader in Asia is routed by DNS to Singapore, which only caches. Most pages are answered right there.
English Wikipedia had about 85.5B page views and 61M edits in 2025. Roughly how many views per edit?
Pick one — or just keep scrolling, it's not graded.
Keep going — it's free
Create your free account to unlock the rest of Wikipedia's system design, keep your progress and pick up right where you left off.
- 2Reading
- 3Editing
Ask anything about this system, or share how you'd design it.
Log in to join the conversation