Updated on 2024-05-20 GMT+08:00

Hash Partitioning

Hash partitioning uses a hash algorithm to map data to partitions based on partition keys. The GaussDB built-in hash algorithm is used. When the value range of partition keys has no data skew, the hash algorithm evenly distributes rows among partitions to ensure that the partition sizes are roughly the same. Therefore, hash partitioning is an ideal method for evenly distributing data among partitions. Hash partitioning is also an easy-to-use alternative to range partitioning, especially when the data to be partitioned is not historical data or has no obvious partition key. The following is an example:

gaussdb=# CREATE TABLE bmsql_order_line (
    ol_w_id          INTEGER   NOT NULL,
    ol_d_id          INTEGER   NOT NULL,
    ol_o_id          INTEGER   NOT NULL,
    ol_number        INTEGER   NOT NULL,
    ol_i_id          INTEGER   NOT NULL,
    ol_delivery_d    TIMESTAMP,
    ol_amount        DECIMAL(6,2),
    ol_supply_w_id   INTEGER,
    ol_quantity      INTEGER,
    ol_dist_info     CHAR(24)
)
-- Define 100 partitions.
PARTITION BY HASH(ol_d_id)
(
    PARTITION p0,
    PARTITION p1,
    PARTITION p2,
    …
    PARTITION p99
);

In the preceding example, the ol_d_id column in the bmsql_order_line table is partitioned. The ol_d_id column is an identifier attribute column and does not distinguish time or a specific dimension. Using the hash partitioning policy to divide a table is an ideal choice. Compared with operations of other partitioning types, when creating partitions, you only need to specify the partition key and the number of partitions on the basis that the partition key does not have too much data skew (one or more values are highly repeated). In addition, data in each partition is evenly distributed, improving usability of partitioned tables.