Overview
FactorZero is a fault-tolerant distributed file system paired with a parallel workload engine. The project focuses on maintaining correctness and availability when nodes or network communication fail.
What I built
- A distributed file system with whole-file caching, primary-backup replication, and heartbeat-based failover.
- A coordinator-worker engine for distributing large-scale primality-testing tasks with recovery.
- Idempotent gRPC retry semantics and Lamport clocks for ordered result aggregation in unreliable network conditions.
Validation
I verified global-state consistency with Chandy-Lamport snapshots and benchmarked throughput under varied failure scenarios.