TidesDB – A persistent key-value store for fast storage

ACID Transactions

Atomic, consistent, isolated (Read Committed), and durable. Transactions support multiple operations across column families. Writers are serialized per column family ensuring atomicity, while COW provides consistency for concurrent readers.

Great Concurrency

Readers don’t block readers or writers. Optimized reader-writer locks per column family for maximum throughput.

LSM-Tree Architecture

Log-structured merge-tree design optimized for write-heavy workloads with efficient read performance through bloom filters and block indices.

Flexible Compression

Support for Snappy, LZ4, and ZSTD compression algorithms. Configure per column family for optimal storage efficiency.

Background Compaction

Automatic parallel compaction with configurable thread pools. Reduces SSTable count and removes tombstones without blocking operations.

Cross-Platform

Native support for Linux, macOS, and Windows with platform abstraction layer for consistent behavior.

Simple API

Clean, intuitive C API with consistent error handling. All functions return 0 on success with detailed error codes.

Production Ready

Write-ahead logging, TTL support, custom comparators, and configurable sync modes for durability guarantees.