Level 4: Reliability, security & cloud · L4.3 Fault tolerance, disaster recovery & multi-region
Bulkheads, cells and shuffle sharding
Limit blast radius with cells and shuffle sharding.
12 min · Pro
What you take away
- A cell is L3.2's bulkhead scaled up: a complete independent copy of the stack serving a subset of customers, so a failure costs 1/N of them.
And 3 more inside the class, in Pro.
How the class is laid out
- The problem
- Cells: the bulkhead, scaled up
- Shuffle sharding: combinations instead of partitions
- Check your understanding