sofastack/sofa-jraft

Features

• Consensus Engines - Provides a core consensus engine that ensures high availability and data consistency via leader election and log replication.
• Distributed Consensus Stores - Provides a distributed consensus store that ensures consistent state across multiple nodes using the Raft algorithm.
• Replicated State Machines - Implements a replicated state machine that ensures consistency by applying an ordered sequence of log entries across nodes.
• Consistent Read Optimizations - Provides optimized read indices to guarantee linearizable reads without the overhead of full log writes.
• Distributed Key-Value Stores - Ships a replicated key-value store designed to maintain a reliable source of truth across a distributed cluster.
• Quorum-Based Commit Protocols - Requires a majority quorum of nodes to agree on log entries before committing them to the state machine.
• Dynamic Cluster Membership Management - Provides a system for managing dynamic node membership and leader transitions using joint consensus.
• Atomic Record Updates - Provides atomic operations like compare-and-put and merge to ensure data integrity during concurrent updates.
• Raft Implementations - Provides a production-grade Java implementation of the Raft consensus algorithm.
• Sharded Consensus Architectures - Supports multi-group consensus sharding to distribute traffic and increase system throughput through independent consensus groups.
• Joint Consensus Protocols - Implements a two-phase joint consensus process to safely transition cluster membership and prevent split-brain scenarios.
• Distributed Leader Election - Coordinates a priority-based leader election process to ensure high availability and single-writer exclusivity.
• Log Replication Strategies - Replicates logs and state snapshots across the network to ensure continuous service availability and rapid node recovery.
• Consensus Log Replication - Copies log entries from a leader to followers using quorum-based consensus to ensure cluster-wide data consistency.
• Replicated Log Management - Manages the synchronization and maintenance of sequential operation logs across distributed nodes to prevent data loss.
• Consensus Log Storage Integrations - Implements high-performance storage interfaces for the Raft consensus log to ensure durability and system recovery.
• Raft Consensus Implementations - Provides a production-grade Java implementation of the Raft consensus protocol for managing replicated logs and leader election.
• Inter-Node Consensus Communication - Manages the network communication of votes and log entries between consensus peers.
• Linearizable Read Validations - Guarantees linearizable reads by validating leader authority through time-bound leases or heartbeats.
• Leader Authority Validations - Validates leader authority through timed leases and heartbeats to guarantee the most recent write is returned during reads.
• Batch Write Buffering - Supports batch write operations across different partitions in parallel to increase data throughput.
• Leader Distribution Balancers - Enables manual distribution of leaders across nodes to ensure an even workload in multi-group environments.
• Custom Log Serialization - Enables custom serialization and deserialization formats for log entries via a service provider interface.
• Distributed Configuration Storage - Provides consistent storage for cluster-wide configuration and critical system metadata.
• Multi-Raft Sharding - Organizes nodes into multiple independent Raft consensus groups to distribute traffic and increase throughput.
• Learner Node Synchronization - Implements non-voting learner nodes that replicate data from the cluster to scale read operations without affecting quorum.
• Snapshot-Based Compactions - Provides snapshot-based log compaction to truncate obsolete logs and accelerate node recovery.
• Read-Only Data Replication - Supports the deployment of non-voting learner nodes to replicate data for read-scaling and backups.
• Fault-Tolerant Coordinators - Provides a fault-tolerant coordination layer for managing critical system configuration using quorum-based consensus.
• Distributed Locks - Implements distributed reentrant locks with lease management to prevent race conditions in multi-node environments.
• Optimistic Pipelining - Increases replication throughput by asynchronously pipelining multiple log entries to followers.
• Leader-Routed Request Handlers - Tracks the current cluster leader and redirects client requests to the correct node for atomic operations.
• Leader Role Transfers - Supports the explicit transfer of leader authority to facilitate node maintenance and load balancing.
• Consensus Node Roles - Initializes consensus nodes with specific roles and storage paths to manage leadership transitions.
• RPC Servers - Establishes an RPC server to handle internal consensus protocol interactions such as elections and heartbeats.
• Component Extensibility Interfaces - Provides standardized interfaces for implementing custom log storage and entry encoding components.
• Fault Tolerance Implementation - Implements architectural resilience to maintain consistency and availability during network partitions and quorum loss.
• Metric and Performance Monitors - Collects high-frequency statistics on log replication, state machine application, and RPC latency.
• Distributed System Verification - Uses fault-injection and network partitioning tests to verify distributed system correctness and consistency.

Star history