activeJune 2026
Distributed LRU Cache
A sharded, replicated caching layer in Java, built to explore consistent hashing and failure recovery.
JavagRPCConsistent hashing
The problem
Most caching tutorials stop at "here's how LRU works." I wanted to build the part that's actually hard: what happens when a shard dies mid-write, and how do you rebalance without a stop-the-world pause.
Current state
Core LRU + consistent hashing ring is done. Working on gRPC-based replication between shard replicas now, with a target of sub-100ms failover.
Open questions
Still deciding between quorum writes and primary-replica for the consistency model — leaning primary-replica for simplicity, accepting the availability trade-off.