tigerbeetle/tigerbeetle

Features

• Distributed Ledgers - A high-performance, fault-tolerant database designed for mission-critical double-entry bookkeeping.
• Ledger Engines - Builds high-performance, double-entry accounting systems that require strict transactional integrity.
• High-Volume - Executes millions of financial operations per second with low latency by utilizing optimized, lock-free concurrency.
• Accounting Engines - Enforces balance invariants, atomic transfers, and idempotent operations for complex financial workflows.
• Accounting Invariants - TigerBeetle validates debit and credit balances within the database to ensure funds are always accounted for and balance limits are respected.
• Transaction Processors - Optimized for millions of financial ledger entries per second using a single-core architecture.
• Financial Transaction Processing - Executes immutable ledger entries that debit one account and credit another, supporting both single-phase and two-phase transactions.
• High Availability Architectures - Maintains continuous operation through automated consensus-based failover and multi-cloud replication.
• Balance Tracking - Maintains running totals of debits and credits for accounts to allow applications to compute balances.
• Persistence & Durability - Replicates data across multiple nodes using a consensus protocol to ensure records survive hardware or network failures.
• Distributed Databases - Provides a consensus-based storage engine that ensures strict serializability, data durability, and high availability.
• Strict Serializability - Processes all transactions sequentially through a single state machine to ensure absolute consistency.
• Two-Phase Commit - Reserves funds in a pending state and resolves them via posting, voiding, or expiration to support complex financial workflows.
• High Availability Clusters - Runs a cluster of replicas to provide high-availability and data safety for financial transactions across multiple nodes.
• Raft Consensus Implementations - Coordinates multiple server replicas using the Raft protocol to ensure strict serializability and high availability.
• Integer-Based Currency - Uses 128-bit integers to represent currency units, ensuring precise calculations by avoiding floating-point arithmetic.
• Immutable - Records all transactions as append-only entries that cannot be erased or modified, ensuring a permanent and auditable history.
• Account Management - Registers and tracks financial accounts, supporting custom metadata and batch operations to maintain accurate balances.
• Append-Only Storage Engines - Persists all financial transactions as an immutable, checksummed, and hash-chained append-only log for auditability.
• Backup and Recovery - Detects and repairs data file corruption across replicas while enforcing safe shutdown protocols.
• Clock Synchronization Protocols - Aggregates physical clock readings from multiple replicas to maintain a consistent wall-clock time across the distributed system.
• Currency Conversion - Links multiple ledger transfers into a single atomic unit to ensure currency conversion occurs as an indivisible transaction.
• Atomic Transfer Chains - Links multiple individual transfers into a single atomic unit to ensure complex sequences of financial events succeed or fail together.
• Replica Recovery - Restores cluster integrity by re-synchronizing data to a replacement node after a permanent hardware or instance failure.
• Atomic Chains - Links multiple ledger events into a single atomic chain that succeeds or fails as a unit.
• Distributed Database Clusters - Deploys and manages high-availability database clusters that ensure data durability and consistency across multiple geographic nodes.
• Multi-Region Deployment Strategies - Deploys cluster nodes across multiple sites to maintain transaction processing capabilities during site failures.
• Rate Limiting - Restricts the frequency of transactions for specific accounts or entities to protect system throughput and prevent resource exhaustion.
• Ledger Querying - Filters account transfers and historical balances based on specified parameters to retrieve precise financial records.
• System Timestamping - Generates unique, immutable, and totally ordered timestamps for all records upon ingestion.
• Commit Protocols - Implements two-phase commit protocols to reserve funds in a pending state before final settlement for atomic financial operations.
• Operation Idempotency - Uses unique identifiers for accounts and transfers to prevent duplicate processing during retries or system failures.
• Accounting Systems - Distributed financial accounting database for high-performance transaction integrity.
• Balance History - Maintains a chronological record of account balances over time for auditing and reporting purposes.
• Fault Domain Isolation - Enforces the use of independent hardware and network infrastructure for each replica to prevent correlated failures.
• Real-Time Data Streaming - Exports transaction logs and state changes to external systems in real-time for auditing and analytical tasks.
• Sequential Execution Engines - Processes all transactions sequentially on a single core to eliminate lock contention and ensure absolute consistency.
• Zero-Copy Mechanisms - Transfers data directly between network buffers and application memory to minimize CPU overhead and maximize transaction throughput.
• Correction Mechanisms - Reverses or adjusts previously committed transactions by issuing compensating entries that maintain ledger history integrity.
• Metrics Exporters - Streams internal performance and status data to external collection agents using standard protocols for real-time observability.
• System Audit Trails - Records immutable, append-only logs of all financial events to provide a verifiable and permanent history.
• Metadata Storage - Provides indexed fields for arbitrary application data to support querying and entity linking for financial accounts.
• Database Drivers - Provides native language drivers to ensure seamless communication with the storage layer and database operations.
• Cluster Upgrades - Updates the database cluster by replacing binaries across replicas without requiring downtime.
• Request Batching - Combines multiple operations into single network calls to increase data throughput and reduce response latency.
• Static Allocation Strategies - Pre-allocates all required memory at startup to avoid runtime garbage collection pauses and ensure predictable performance.
• Expiration Logic - Automatically releases reserved funds back to original accounts if a transaction is not settled within a predefined time interval.
• Latency Monitoring - Records the duration of specific transaction operations to help evaluate performance and identify bottlenecks within the database engine.
• Fee Recording - Appends a distinct transfer to an exchange chain to capture the spread as a transparent, auditable fee.
• Cache Efficiency Analysis - Exposes hit and miss counts to determine if memory allocation for data caching is appropriately sized for the current workload.
• Compatibility Migrators - Ensures that ledger data files created by older versions remain readable and functional after software upgrades.
• Timestamping Mechanisms - Generates unique and totally ordered timestamps for all records upon ingestion to ensure consistent temporal ordering.
• Storage Configurations - Optimizes data persistence by supporting local NVMe drives, raw block devices, and remote block storage configurations.
• Client Session Management - Registers unique client identifiers with the cluster to establish request sequences and enforce concurrency limits.
• Memory Locking - Prevents the operating system from swapping active memory to disk by configuring system limits to maintain consistent performance.
• Request Lifecycle Control - Controls the execution of database operations through client-side cancellation and connection management for long-running tasks.

Star history