Reliability / 1 / 4
Redundancy and failover
Every part fails sooner or later. How to find single points of failure, add a spare for each, switch over automatically, and not lose data on the way.
- 14 min
- Intermediate
- 12 steps
- Errors
- 100 %
- Time to recover
- 1 800 s
- Lost bookings
- 0
- Write p95
- 12 ms
- request
- reply
- redelivery
- duplicate
- lost
Numbers are a model scenario, not measurements
The real request
Guest → App 1 · zone ARequest
GET /bookings/b_311 HTTP/1.1
Host: book.example.com
Response
connect to 203.0.113.10:443: Connection refused1
Connection refused— Nothing listens on the port. The guest gets an error page, the hotel gets no booking.
Course: 42 lessons
01 Foundations
02 Scaling
03 Data & storage
04 Communication
05 Reliability
06 Architecture
07 Security
08 Shipping
09 Whole systems
10 Business processes