spacejam/sled

Features

• Key-Value - Functions as a high-performance embedded key-value store supporting atomic updates and conditional modifications.
• ACID Transactional Cores - Provides an ACID-compliant core for atomic and isolated multi-key transaction processing.
• Database Atomic Batches - Groups multiple database mutations into a single atomic operation to ensure consistency.
• Write Persistence Guarantees - Guarantees data persistence by flushing dirty IO buffers to disk synchronously or asynchronously.
• Zero-Copy Memory Mappings - Provides direct references to cached values using epoch-based reclamation to eliminate memory allocations during reads.
• High-Performance Persistence Layers - Writes structured binary data to stable storage with specific optimizations for flash memory and crash recovery.
• Key-Value Stores - Provides a high-performance, disk-backed embedded key-value store for byte arrays.
• B-Tree - Implements a B+ Tree indexing structure to enable efficient range-based scans and fast key lookups.
• Log-Structured Merge-Trees - Utilizes a log-structured merge-tree architecture to optimize write throughput and read performance.
• Cursor-based Iterators - Implements cursor-based iterators for memory-efficient record-by-record processing of the database.
• Strict Serializability - Ensures serializable transaction isolation for atomic multi-key updates across different data trees.
• Strict Serializability - Ensures absolute consistency through serializable transactions across multiple data trees.
• Log State Recovery - Reconstructs database state by applying log segments to snapshots for reliable crash recovery.
• Write-Ahead Logging - Uses write-ahead logging to record every mutation to a persistent log for guaranteed crash recovery.
• Log-Structured Storage Engines - Implements a log-structured storage engine using write-ahead logging and page caches for high write throughput.
• Zero-Copy Data Access - Provides zero-copy data access by returning direct references to cached values using epoch-based reclamation.
• Database Page Caching - Implements database page caching to store data fragments in memory and reduce write amplification on flash storage.
• Storage Key Observation - Monitors specific key prefixes to trigger reactive events when records are updated.
• Conditional Updates - Allows modifying or deleting a value only if the current value matches a specific expected state.
• Keyspace Isolation - Supports creating and removing disk-backed trees to isolate different keyspaces within a single database.
• Epoch-Based Reclamation - Uses epoch-based reclamation to provide zero-copy reads without immediate memory allocations.
• Data Replication - Implements binary stream replication to synchronize data and tree updates across distributed nodes.
• Atomic Key-Value Operations - Supports atomic key-value operations that execute read-modify-write cycles as single, indivisible units.
• Disk Defragmentation - Reclaims disk space and reduces fragmentation by moving live pages from underutilized segments.
• Change Notifications - Provides reactive data observation via key-prefix event subscriptions to trigger actions when records are updated.
• Key Deletion - Provides capabilities for removing values from the database by their unique keys.
• Custom Value Merging - Provides a custom merge operator to execute read-modify-write logic directly during the write process.
• Record Compression - Reduces the storage footprint of the database by applying high-efficiency compression to stored records.
• Storage Space Optimization - Implements disk defragmentation and page cache optimizations to reduce write amplification on flash storage.
• Bases de données - Embedded database with lock-free concurrency.

Historique des stars