Badger vs RocksDB: What are the differences?

# Introduction

Key differences between Badger and RocksDB:

1. **Storage Engine**: Badger uses a log-structured merge-tree (LSM) as its storage engine, while RocksDB uses a variant of LSM called adjustable-size LSM (ASLSM) or also known as Universal Compaction. The main difference lies in how they manage data compaction and merging of sorted data.
   
2. **Memory Usage**: Badger uses a pure in-memory structure for caching frequently accessed data, resulting in a higher memory footprint compared to RocksDB, which relies on a combination of in-memory and on-disk caching mechanisms to manage memory consumption efficiently.
   
3. **Concurrency Control**: RocksDB employs optimistic, fine-grained locking at the column-family level to allow for concurrent reads and writes, providing better performance in read-heavy workloads. In contrast, Badger uses a simpler locking mechanism at the key level, which may lead to reduced concurrency in certain scenarios.
   
4. **Coding Language**: Badger is written in Go, making it easier for developers familiar with the language to contribute and maintain the codebase. RocksDB, on the other hand, is implemented in C++, which may require a different skill set to work with and understand.
   
5. **Crash Recovery**: Badger supports crash recovery by utilizing simple write-ahead logging (WAL) mechanisms to ensure durability, while RocksDB offers a more granular commit logs approach that enables faster recovery times in case of system failures.
   
6. **Performance Tuning**: RocksDB provides more configuration options for fine-tuning performance parameters such as compaction strategies, block size, and bloom filters, giving users greater control over optimizing their database performance compared to Badger's more simplistic configuration settings.

In Summary, Badger and RocksDB differ in their storage engine type, memory usage, concurrency control, coding language, crash recovery mechanism, and performance tuning options.