Glosarium AI
Kamus lengkap Kecerdasan Buatan
Synchronous Replication
Replication mechanism where writes are confirmed only after being validated on all replicated nodes, guaranteeing strong consistency but increasing latency.
Asynchronous Replication
Replication strategy where writes are immediately confirmed on the primary node before propagation to replicas, offering better performance at the expense of immediate consistency.
Vector Clocks
Data structure used to track causal relationships between events in a distributed system to detect concurrency conflicts.
Leader Election
Algorithmic process allowing a cluster of nodes to dynamically designate a unique leader node to coordinate operations and avoid split-brain scenarios.
Heartbeat
Periodic signal sent by nodes to indicate their operational status, enabling rapid failure detection in distributed systems.
Raft
Understandable distributed consensus algorithm that guarantees data consistency through leader election and log replication in clusters.
Paxos
Fundamental consensus algorithm allowing faulty nodes to agree on a single value in an asynchronous distributed environment.
Multi-Master Replication
Replication architecture where multiple nodes can accept simultaneous writes, requiring complex conflict resolution mechanisms.
Read Replicas
Read-only copies of the primary database used to distribute read load and improve overall system availability.
Conflict Resolution
Set of strategies (LWW, CRDTs, manual merge) to resolve inconsistencies when concurrent modifications are detected between replicas.
Split Brain
Critical condition where a cluster splits into isolated sub-networks that can each elect a leader, risking serious data inconsistencies.
Distributed Transaction
Atomic transaction spanning multiple nodes or databases, requiring protocols like Two-Phase Commit to guarantee integrity.