datastore/tests/performance/limit_by_with_order.xml

<test>
    <substitutions>
        <substitution>
            <name>cardinality</name>
            <values>
                <value>1000</value>
                <value>100000</value>
            </values>
        </substitution>
    </substitutions>

    <create_query>
        CREATE TABLE test_limit_by_{cardinality}
        (a UInt32, b UInt32, v UInt64)
        ENGINE = MergeTree
        ORDER BY (a, b)
    </create_query>

    <fill_query>
        INSERT INTO test_limit_by_{cardinality}
        SELECT number % {cardinality}, number % 97, number
        FROM numbers(5000000)
    </fill_query>

    <fill_query>
        INSERT INTO test_limit_by_{cardinality}
        SELECT number % {cardinality}, number % 97, number
        FROM numbers(5000000)
    </fill_query>

    <!-- Full sort, unordered BY (BY column is not a sort prefix) -->
    <query>
        SELECT number % {cardinality} AS k, number AS v
        FROM numbers_mt(10000000)
        ORDER BY v LIMIT 10 BY k
        FORMAT NULL
    </query>

    <!-- Full sort (BY column is the sort prefix) -->
    <query>
        SELECT number % {cardinality} AS k, number AS v
        FROM numbers_mt(10000000)
        ORDER BY k, v LIMIT 10 BY k
        FORMAT NULL
    </query>

    <!-- Full sort with a derived BY expression -->
    <query>
        SELECT number % {cardinality} AS grp, number AS v
        FROM numbers_mt(10000000)
        ORDER BY v LIMIT 10 BY grp % 64
        FORMAT NULL
    </query>

    <!-- Full sort with an expensive BY expression -->
    <query>
        SELECT number, concat(toString(number), '-', lowerUTF8(toString(number * 31)), '-', toString(number % 97)) AS bucket
        FROM numbers_mt(10000000)
        ORDER BY number LIMIT 10 BY bucket
        FORMAT NULL
    </query>

    <!-- Read-in-order from MergeTree -->
    <query>
        SELECT a, b, v FROM test_limit_by_{cardinality}
        ORDER BY a, b LIMIT 10 BY a
        FORMAT NULL
    </query>

    <!-- Read-in-order from MergeTree, unordered BY -->
    <query>
        SELECT a, b, v FROM test_limit_by_{cardinality}
        ORDER BY a, b LIMIT 10 BY b
        FORMAT NULL
    </query>

    <!-- Read-in-order with a trailing non-PK sort column (prefix-sorted input) -->
    <query>
        SELECT a, b, v FROM test_limit_by_{cardinality}
        ORDER BY a, b, v + 1 LIMIT 10 BY a
        FORMAT NULL
    </query>

    <!-- Skewed key distribution: ~90% of rows fall into a single hot key, interleaved with cold keys -->
    <query>
        SELECT if(number % 10 = 0, number % {cardinality}, 0) AS k, number AS v
        FROM numbers_mt(10000000)
        ORDER BY k, v LIMIT 10 BY k
        FORMAT NULL
    </query>

    <!-- UInt8 BY key: narrow integer type -->
    <query>
        SELECT toUInt8(number % {cardinality}) AS k, number AS v
        FROM numbers_mt(10000000)
        ORDER BY k, v LIMIT 10 BY k
        FORMAT NULL
    </query>

    <!-- LowCardinality(String) BY key -->
    <query>
        SELECT toLowCardinality(toString(number % {cardinality})) AS k, number AS v
        FROM numbers_mt(10000000)
        ORDER BY k, v LIMIT 10 BY k
        FORMAT NULL
    </query>

    <!-- Two String BY columns -->
    <query>
        SELECT toString(number % {cardinality}) AS k1, toString(number % 97) AS k2, number AS v
        FROM numbers_mt(10000000)
        ORDER BY k1, k2, v LIMIT 10 BY k1, k2
        FORMAT NULL
    </query>

    <drop_query>DROP TABLE IF EXISTS test_limit_by_{cardinality}</drop_query>
</test>