DBRaven
Partitioning

Partition Hotspots

Intermediate

How partition key design determines load distribution, how sequential keys create hotspot partitions, why hotspots cause cascading failures, and how to design partition keys that distribute load uniformly.

Step 1 of 5

Uniform Distribution: Ideal Partitioning

A well-designed partition key distributes writes uniformly across all partitions. Each partition receives roughly equal throughput, memory, and disk usage. In this state, adding capacity scales horizontally: more nodes, more partitions, proportionally more throughput. This is the baseline partitioned system should target.

P0

0x0000–0x1FFF

18%

P1

0x2000–0x3FFF

22%

P2

0x4000–0x5FFF

19%

P3

0x6000–0x7FFF

21%

P4

0x8000–0x9FFF

20%

P5

0xA000–0xBFFF

18%

P6

0xC000–0xDFFF

21%

P7

0xE000–0xFFFF

21%

Uniform distribution: load within 5% variance across all 8 partitions

Key Takeaways

  • Uniform distribution is the prerequisite for horizontal scale-out to actually work
  • Consistent hashing distributes by token range: the partition key's hash determines placement
1 / 5