Hot Spots

Summary: A partition with a disproportionately high load compared to other partitions in a cluster.

Sources: chapter6

Last updated: 2026-04-15


If partitioning is unfair, so that some partitions have more data or queries than others, it is called skewed. A partition with disproportionately high load is called a hot spot. (source: chapter6)

Causes

  • Skewed Data: Uneven distribution of keys.
  • Access Patterns: For example, if the key is a timestamp, all writes for “today” might go to the same partition (in key-range-partitioning). (source: chapter6)
  • High Traffic Keys: A single key being very popular (e.g., a celebrity on a social media site). (source: chapter6)

Mitigation

  • Using hash-partitioning to distribute keys more randomly.
  • Adding a random number to the beginning or end of a very “hot” key to split it across multiple partitions (requires application-level bookkeeping). (source: chapter6)