distributed

**Distributed Hash Table DHT Parallel** is **a decentralized data structure storing key-value pairs across network of machines without central server, enabling scalable, fault-tolerant storage and retrieval** — foundational for peer-to-peer systems, distributed caching, and content distribution. DHTs provide hash table semantics at distributed scale. **Consistent Hashing** maps keys and nodes to same hash space (e.g., 0 to 2^160 arranged on ring). Each key belongs to closest node clockwise on ring. Advantage: adding/removing nodes affects only O(K/N) keys (K=total keys, N=nodes), unlike traditional hashing where nearly all keys rehash. Node joins: new node claims keys between it and predecessor. **Chord Protocol** implements DHT with O(log N) lookup hops: each node maintains finger table with N entries, entry i points to node at position (node_id + 2^(i-1)) mod 2^160. Lookup starts at closest preceding node, finger table guides toward target with logarithmic hops. Node joins: new node discovers predecessor/successor via lookup, updates affected finger tables. **Kademlia Protocol** uses XOR metric for distance: closer nodes have more shared bit prefixes. Lookup is faster in practice than Chord—splits remaining target bits in half each step (ternary search). Nodes maintain buckets of peers at each distance range, enabling parallel lookups to multiple peers. **Replication and Fault Tolerance** store key replicas on successor nodes (e.g., next k nodes after primary owner). Failure of single node doesn't lose data. **Parallel Lookups** send queries to multiple peers simultaneously, first response returns data. Latency reduced from sequential hops to single parallel round. **Load Balancing** virtual nodes distribute load—single physical machine hosts multiple virtual nodes with different hash IDs, spreading keys more evenly. **Consistency Models** eventual consistency accepts temporary inconsistency: updates propagate asynchronously. Strong consistency requires global coordination, reducing performance. Quorum reads/writes balance consistency and availability. **Applications** include IPFS distributed file system, BitTorrent peer discovery, blockchain consensus, and distributed cache. **Effective DHT design balances lookup latency, replication overhead, and consistency guarantees** for specific application requirements.