Rocksdb compact调优参数

CompactionStyle

enum CompactionStyle : char {
  // level based compaction style
  kCompactionStyleLevel = 0x0,
  // Universal compaction style
  // Not supported in ROCKSDB_LITE.
  kCompactionStyleUniversal = 0x1,
  // FIFO compaction style
  // Not supported in ROCKSDB_LITE
  kCompactionStyleFIFO = 0x2,
  // Disable background compaction. Compaction jobs are submitted
  // via CompactFiles().
  // Not supported in ROCKSDB_LITE
  kCompactionStyleNone = 0x3,
};

rocksdb 有三种compact style：

• kCompactionStyleLevel：rocksdb默认的compact style。这种策略保证每层的key都是有序并且互不重叠，但层于层直接的key可能会重叠。当Ln层满足compact条件时，会根据CompactionPri策略（下面会介绍），挑选出Ln层的文件file1和Ln+1层中与file1有key重叠的file进行合并，生成file2，并将file2插入到Ln+1层中。该策略拥有最小的空间放大，但是带来很大的写放大和读放大。
• kCompactionStyleUniversal：这种style将磁盘上数据完全按照写入时间线去组织，只将相邻时间段内写入磁盘上的数据compact，很像在合并互不重叠(overlapping)的时间段。该策略写放大比较低，但是读放大和空间放大比较高。
• kCompactionStyleFIFO：通常用于几乎不修改的场景，比如消息队列、时序数据库等。
• kCompactionStyleNone：关闭自动compact，手动触发执行。

CompactionPri

// In Level-based compaction, it Determines which file from a level to be
// picked to merge to the next level. We suggest people try
// kMinOverlappingRatio first when you tune your database.
enum CompactionPri : char {
  // Slightly prioritize larger files by size compensated by #deletes
  kByCompensatedSize = 0x0,
  // First compact files whose data's latest update time is oldest.
  // Try this if you only update some hot keys in small ranges.
  kOldestLargestSeqFirst = 0x1,
  // First compact files whose range hasn't been compacted to the next level
  // for the longest. If your updates are random across the key space,
  // write amplification is slightly better with this option.
  kOldestSmallestSeqFirst = 0x2,
  // First compact files whose ratio between overlapping size in next level
  // and its size is the smallest. It in many cases can optimize write
  // amplification.
  kMinOverlappingRatio = 0x3,
};

该参数控制每次compact，挑选哪些文件参与compact。

• kByCompensatedSize：优先选择被删除数据最多的file去compact。rocksdb的默认设置。被删除的数据越早被compact，越可以减少空间放大，同时降低读放大。
• kOldestLargestSeqFirst：优先选择冷数据file去compact。将冷数据compact到下一层，而将热数据保留在上层，可以有效降低读放大问题。适用有冷热数据之分的场景。
• kOldestSmallestSeqFirst：优先选择key范围密集的file去compact。file中key范围越密集，就会有更少的重叠file在下一level，从而降低compact开销。适用于key空间均匀分布的场景。
• kMinOverlappingRatio：优先选择和下层key范围重叠率最小的文件去compact。可以降低写放大。

参考资料

• https://rocksdb.org/blog/2016/01/29/compaction_pri.html(https://rocksdb.org/blog/2016/01/29/compaction_pri.html)