Partitioning

Distribution

Distribution counts the keys routed to each partition.

highlighted = computed this step

Distribution is counted exactly

Once every key is routed, distribution is just exact per-partition counting. The count table is not sampled or estimated. Note: the key lists in the render explain each count.

route then count\text{route then count}

Range distribution

The range scheme has 4 partitions and maximum partition count 3.

range partitions=4,max count=3\text{range partitions}=4,\quad \text{max count}=3

Range and hash sharding are exact routing functions on tiny pinned key sets; real sharding adds consistent hashing, skew, hot partitions, and rebalancing cost, which are empirical and operational concerns beyond this model - no product claims.

Range distribution - range:10,20,30partitioncountkeysP032,4,6P1112P2122P3232,42

Hash distribution

The hash scheme has 3 partitions and maximum partition count 3. Note: these counts come from explicit remainder routing.

hash partitions=3,max count=3\text{hash partitions}=3,\quad \text{max count}=3

Range and hash sharding are exact routing functions on tiny pinned key sets; real sharding adds consistent hashing, skew, hot partitions, and rebalancing cost, which are empirical and operational concerns beyond this model - no product claims.

Hash distribution - hash:mod 3partitioncountkeysP023,6P131,4,7P222,5

Summary

Distribution is a recomputed table of counts and routed key lists. Note: no model here predicts operational skew or throughput.

distribution is exact\text{distribution is exact}