This project implements a fault-tolerant, sharded key/value storage system using the Raft consensus algorithm for replication and consistency.
The system partitions its keyspace into shards, each managed by a replica group, and a replicated Shard Controller dynamically assigns shards to groups.
The goal is to build a scalable, reconfigurable storage system capable of maintaining strong consistency even during failures or reconfiguration events.

• Manages a sequence of configurations mapping shards to replica groups.
• Handles the following RPCs:
  • Join: Add new replica groups
  • Leave: Remove replica groups
  • Move: Manually reassign a shard to a different group
  • Query: Fetch the latest or specific configuration
• Uses Raft for replication and fault-tolerance.
• Balances shards evenly while minimizing unnecessary data movement.

• Each replica group serves Get/Put/Append requests for its assigned shards.
• Uses Raft to replicate operations and maintain strong consistency.
• Handles shard migration during configuration changes:
  • Transfers data via RPC between groups.
  • Rejects requests for shards it no longer owns.
• Guarantees:
  • Linearizability (operations appear in a single global order).
  • At-most-once semantics for client requests.
• Functions correctly under partial failures or network partitions.

• Dynamic Shard Reconfiguration — automatic reassignment of shards on joins/leaves.
• Fault-Tolerant Coordination — Raft ensures consistent replication across all groups.
• Shard Migration — seamless and consistent data transfer during reconfiguration.
• Duplicate Request Handling — idempotence across unreliable network conditions.