Different Partitioning Strategies in Databases

This section will discuss two strategies to partition the dataset in a distributed environment:

• Partition by a key range

• Partition by key hash

Partition by a key range

In this strategy, we divide a continuous range of keys into buckets. Then, we assign each bucket to a partition. A single host instance can hold multiple partitions. The range of keys assigned to a bucket may or may not be continuous. Within each partition, they store keys in sorted order, thus facilitating range scan queries.

In the example above:

• We have 3 host instances namely Node 1, Node 2, and Node 3.

• Node 1 has 2 partitions.

  • Partition 1 includes keys starting from A, B, C, D, E.

  • Partition 2 includes keys starting from F, G, H, I, J.

• Node 2 has 2 partitions.

  • Partition 3 includes keys starting from K , L , M, N, O.

  • Partition 4 includes keys starting from P, Q, ...