datastore/tests/performance/limit_by_partitions.xml

<test>
    <substitutions>
        <substitution>
            <name>size</name>
            <values>
                <value>100000</value>
                <value>10000000</value>
            </values>
        </substitution>

        <substitution>
            <name>partitions</name>
            <values>
                <value>16</value>
                <value>64</value>
            </values>
        </substitution>
    </substitutions>

    <create_query>CREATE TABLE t_{size}_{partitions} (a UInt64, b UInt64) ENGINE = MergeTree ORDER BY a PARTITION BY sipHash64(a) % {partitions}</create_query>

    <!-- Skewed shape: 3 hot values × 30% rows each (90% total) + 27 values × ~0.37% each (10% total). -->
    <create_query>CREATE TABLE t_skew_{size}_{partitions} (a UInt64, b UInt64) ENGINE = MergeTree ORDER BY tuple() PARTITION BY a</create_query>

    <fill_query>INSERT INTO t_{size}_{partitions} SELECT number, number FROM numbers_mt({size})</fill_query>
    <fill_query>OPTIMIZE TABLE t_{size}_{partitions} FINAL</fill_query>

    <fill_query>INSERT INTO t_skew_{size}_{partitions} SELECT if(number % 10 &lt; 9, number % 3, 3 + (number % 27)) AS a, number AS b FROM numbers_mt({size})</fill_query>
    <fill_query>OPTIMIZE TABLE t_skew_{size}_{partitions} FINAL</fill_query>

    <query>SELECT a FROM t_{size}_{partitions} LIMIT 10 BY a FORMAT Null</query>

    <query>SELECT a FROM t_{size}_{partitions} WHERE b > 1 LIMIT 10 BY a FORMAT Null</query>

    <query>SELECT a, b FROM t_{size}_{partitions} LIMIT 10 BY a, b FORMAT Null</query>

    <query>SELECT a FROM t_{size}_{partitions} LIMIT 100000000 BY a FORMAT Null</query>

    <query>SELECT a FROM t_{size}_{partitions} ORDER BY b DESC LIMIT 10 BY a FORMAT Null</query>

    <query>SELECT a FROM t_skew_{size}_{partitions} LIMIT 10 BY a FORMAT Null</query>

    <drop_query>DROP TABLE t_{size}_{partitions}</drop_query>
    <drop_query>DROP TABLE t_skew_{size}_{partitions}</drop_query>
</test>