从 leveldb 到分布式 kv 存储系统

LevelDB

详细原理见 leveldb 原理解析

RocksDB : enhanced LevelDB

https://rocksdb.org.cn/doc.html
RocksDB是从 levelDB 的某个 commit checkout 出来的，进化版

RocksDB is a C++ library providing an embedded key-value store, where keys and values are arbitrary byte streams. It was developed at Facebook based on LevelDB and provides backwards-compatible support for LevelDB APIs.

RocksDB is optimized for Flash with extremely low latencies. RocksDB uses a Log Structured Database Engine for storage, written entirely in C++. A Java version called RocksJava is currently in development. See RocksJava Basics.

RocksDB features highly flexible configuration settings that may be tuned to run on a variety of production environments, including pure memory, Flash, hard disks or HDFS. It supports various compression algorithms and good tools for production support and debugging.

Features

1. Designed for application servers wanting to store up to a few terabytes of data on locally attached Flash drives or in RAM
2. Optimized for storing small to medium size key-values on fast storage – flash devices or in-memory
3. Scales linearly with number of CPUs so that it works well on processors with many cores

RocksDB Features that are not in LevelDB

Performance

1. Multithread compaction
2. Multithread memtable inserts
3. Reduced DB mutex holding
4. Optimized level-based compaction style and universal compaction style
5. Prefix bloom filter
6. Memtable bloom filter
7. Single bloom filter covering the whole SST file
8. Write lock optimization
9. Improved Iter::Prev() performance
10. Fewer comparator calls during SkipList searches
11. Allocate memtable memory using huge page.

Features
12. Column Families
13. Transactions and WriteBatchWithIndex
14. Backup and Checkpoints