Examples with MergeJoinPOP org.apache.drill.exec.physical.config.MergeJoinPOP used on opensource projects

Example 11 with MergeJoinPOP

use of org.apache.drill.exec.physical.config.MergeJoinPOP in project drill by axbaretto.

the class TestOutputBatchSize method testMergeJoinUpperLimit.

@Test
public void testMergeJoinUpperLimit() throws Exception {
    // test the upper limit of 65535 records per batch.
    MergeJoinPOP mergeJoin = new MergeJoinPOP(null, null, Lists.newArrayList(joinCond("c1", "EQUALS", "c2")), JoinRelType.LEFT);
    mockOpContext(mergeJoin, initReservation, maxAllocation);
    numRows = 100000;
    // create left input rows like this.
    // "a1" : 5,  "c1" : <id>
    List<String> leftJsonBatches = Lists.newArrayList();
    StringBuilder leftBatchString = new StringBuilder();
    leftBatchString.append("[");
    for (int i = 0; i < numRows; i++) {
        leftBatchString.append("{\"a1\": 5, " + "\"c1\" : " + i + "},");
    }
    leftBatchString.append("{\"a1\": 5, " + "\"c1\" : " + numRows + "}");
    leftBatchString.append("]");
    leftJsonBatches.add(leftBatchString.toString());
    // create right input rows like this.
    // "a2" : 6, "c2" : <id>
    List<String> rightJsonBatches = Lists.newArrayList();
    StringBuilder rightBatchString = new StringBuilder();
    rightBatchString.append("[");
    for (int i = 0; i < numRows; i++) {
        rightBatchString.append("{\"a2\": 6, " + "\"c2\" : " + i + "},");
    }
    rightBatchString.append("{\"a2\": 6, " + "\"c2\" : " + numRows + "}");
    rightBatchString.append("]");
    rightJsonBatches.add(rightBatchString.toString());
    // output rows will be like this.
    // "a1" : 5,  "c1" : 1, "a2":6,  "c2": 1
    // "a1" : 5,  "c1" : 2, "a2":6,  "c2": 2
    // "a1" : 5,  "c1" : 3, "a2":6,  "c2": 3
    // expect two batches, batch limited by 65535 records
    OperatorTestBuilder opTestBuilder = opTestBuilder().physicalOperator(mergeJoin).baselineColumns("a1", "c1", "a2", "c2").expectedNumBatches(// verify number of batches
    2).inputDataStreamsJson(Lists.newArrayList(leftJsonBatches, rightJsonBatches));
    for (long i = 0; i < numRows + 1; i++) {
        opTestBuilder.baselineValues(5l, i, 6l, i);
    }
    opTestBuilder.go();
}

Example 12 with MergeJoinPOP

use of org.apache.drill.exec.physical.config.MergeJoinPOP in project drill by axbaretto.

the class TestOutputBatchSize method testMergeJoinSingleOutputBatch.

@Test
public void testMergeJoinSingleOutputBatch() throws Exception {
    MergeJoinPOP mergeJoin = new MergeJoinPOP(null, null, Lists.newArrayList(joinCond("c1", "EQUALS", "c2")), JoinRelType.INNER);
    mockOpContext(mergeJoin, initReservation, maxAllocation);
    // create multiple batches from both sides.
    numRows = 4096 * 2;
    // create left input rows like this.
    // "a1" : 5, "b1" : wideString, "c1" : <id>
    List<String> leftJsonBatches = Lists.newArrayList();
    StringBuilder leftBatchString = new StringBuilder();
    leftBatchString.append("[");
    for (int i = 0; i < numRows; i++) {
        leftBatchString.append("{\"a1\": 5, " + "\"b1\" : " + "\"" + wideString + "\"," + "\"c1\" : " + i + "},");
    }
    leftBatchString.append("{\"a1\": 5, " + "\"b1\" : " + "\"" + wideString + "\"," + "\"c1\" : " + numRows + "}");
    leftBatchString.append("]");
    leftJsonBatches.add(leftBatchString.toString());
    // create right input rows like this.
    // "a2" : 6, "b2" : wideString, "c2" : <id>
    List<String> rightJsonBatches = Lists.newArrayList();
    StringBuilder rightBatchString = new StringBuilder();
    rightBatchString.append("[");
    for (int i = 0; i < numRows; i++) {
        rightBatchString.append("{\"a2\": 6, " + "\"b2\" : " + "\"" + wideString + "\"," + "\"c2\" : " + i + "},");
    }
    rightBatchString.append("{\"a2\": 6, " + "\"b2\" : " + "\"" + wideString + "\"," + "\"c2\" : " + numRows + "}");
    rightBatchString.append("]");
    rightJsonBatches.add(rightBatchString.toString());
    // output rows will be like this.
    // "a1" : 5, "b1" : wideString, "c1" : 1, "a2":6, "b2" : wideString, "c2": 1
    // "a1" : 5, "b1" : wideString, "c1" : 2, "a2":6, "b2" : wideString, "c2": 2
    // "a1" : 5, "b1" : wideString, "c1" : 3, "a2":6, "b2" : wideString, "c2": 3
    List<String> expectedJsonBatches = Lists.newArrayList();
    StringBuilder expectedBatchString = new StringBuilder();
    expectedBatchString.append("[");
    for (int i = 0; i < numRows; i++) {
        expectedBatchString.append("{\"a1\": 5, " + "\"b1\" : " + "\"" + wideString + "\"," + "\"c1\" : " + i);
        expectedBatchString.append(", \"a2\": 6, " + "\"b2\" : " + "\"" + wideString + "\"," + "\"c2\" : " + i + "},");
    }
    expectedBatchString.append("{\"a1\": 5, " + "\"b1\" : " + "\"" + wideString + "\"," + "\"c1\" : " + numRows);
    expectedBatchString.append(", \"a2\": 6, " + "\"b2\" : " + "\"" + wideString + "\"," + "\"c2\" : " + numRows + "}");
    expectedBatchString.append("]");
    expectedJsonBatches.add(expectedBatchString.toString());
    long totalSize = getExpectedSize(expectedJsonBatches);
    // set the output batch size to twice of total size expected.
    // We should get 1 batch.
    fragContext.getOptions().setLocalOption("drill.exec.memory.operator.output_batch_size", totalSize * 2);
    OperatorTestBuilder opTestBuilder = opTestBuilder().physicalOperator(mergeJoin).baselineColumns("a1", "b1", "c1", "a2", "b2", "c2").expectedNumBatches(// verify number of batches
    1).expectedBatchSize(// verify batch size
    totalSize).inputDataStreamsJson(Lists.newArrayList(leftJsonBatches, rightJsonBatches));
    for (long i = 0; i < numRows + 1; i++) {
        opTestBuilder.baselineValues(5l, wideString, i, 6l, wideString, i);
    }
    opTestBuilder.go();
}

Example 13 with MergeJoinPOP

use of org.apache.drill.exec.physical.config.MergeJoinPOP in project drill by axbaretto.

the class TestOutputBatchSize method testMergeJoinMultipleOutputBatches.

@Test
public void testMergeJoinMultipleOutputBatches() throws Exception {
    MergeJoinPOP mergeJoin = new MergeJoinPOP(null, null, Lists.newArrayList(joinCond("c1", "EQUALS", "c2")), JoinRelType.INNER);
    mockOpContext(mergeJoin, initReservation, maxAllocation);
    // create left input rows like this.
    // "a1" : 5, "b1" : wideString, "c1" : <id>
    List<String> leftJsonBatches = Lists.newArrayList();
    StringBuilder leftBatchString = new StringBuilder();
    leftBatchString.append("[");
    for (int i = 0; i < numRows; i++) {
        leftBatchString.append("{\"a1\": 5, " + "\"b1\" : " + "\"" + wideString + "\"," + "\"c1\" : " + i + "},");
    }
    leftBatchString.append("{\"a1\": 5, " + "\"b1\" : " + "\"" + wideString + "\"," + "\"c1\" : " + numRows + "}");
    leftBatchString.append("]");
    leftJsonBatches.add(leftBatchString.toString());
    // create right input rows like this.
    // "a2" : 6, "b2" : wideString, "c2" : <id>
    List<String> rightJsonBatches = Lists.newArrayList();
    StringBuilder rightBatchString = new StringBuilder();
    rightBatchString.append("[");
    for (int i = 0; i < numRows; i++) {
        rightBatchString.append("{\"a2\": 6, " + "\"b2\" : " + "\"" + wideString + "\"," + "\"c2\" : " + i + "},");
    }
    rightBatchString.append("{\"a2\": 6, " + "\"b2\" : " + "\"" + wideString + "\"," + "\"c2\" : " + numRows + "}");
    rightBatchString.append("]");
    rightJsonBatches.add(rightBatchString.toString());
    // output rows will be like this.
    // "a1" : 5, "b1" : wideString, "c1" : 1, "a2":6, "b2" : wideString, "c2": 1
    // "a1" : 5, "b1" : wideString, "c1" : 2, "a2":6, "b2" : wideString, "c2": 2
    // "a1" : 5, "b1" : wideString, "c1" : 3, "a2":6, "b2" : wideString, "c2": 3
    List<String> expectedJsonBatches = Lists.newArrayList();
    StringBuilder expectedBatchString = new StringBuilder();
    expectedBatchString.append("[");
    for (int i = 0; i < numRows; i++) {
        expectedBatchString.append("{\"a1\": 5, " + "\"b1\" : " + "\"" + wideString + "\"," + "\"c1\" : " + i);
        expectedBatchString.append(", \"a2\": 6, " + "\"b2\" : " + "\"" + wideString + "\"," + "\"c2\" : " + i + "},");
    }
    expectedBatchString.append("{\"a1\": 5, " + "\"b1\" : " + "\"" + wideString + "\"," + "\"c1\" : " + numRows);
    expectedBatchString.append(", \"a2\": 6, " + "\"b2\" : " + "\"" + wideString + "\"," + "\"c2\" : " + numRows + "}");
    expectedBatchString.append("]");
    expectedJsonBatches.add(expectedBatchString.toString());
    long totalSize = getExpectedSize(expectedJsonBatches);
    // set the output batch size to 1/2 of total size expected.
    // We will get approximately 4 batches because of fragmentation factor of 2 accounted for in merge join.
    fragContext.getOptions().setLocalOption("drill.exec.memory.operator.output_batch_size", totalSize / 2);
    OperatorTestBuilder opTestBuilder = opTestBuilder().physicalOperator(mergeJoin).baselineColumns("a1", "b1", "c1", "a2", "b2", "c2").expectedNumBatches(// verify number of batches
    4).expectedBatchSize(// verify batch size
    totalSize / 2).inputDataStreamsJson(Lists.newArrayList(leftJsonBatches, rightJsonBatches));
    for (long i = 0; i < numRows + 1; i++) {
        opTestBuilder.baselineValues(5l, wideString, i, 6l, wideString, i);
    }
    opTestBuilder.go();
}

Example 14 with MergeJoinPOP

use of org.apache.drill.exec.physical.config.MergeJoinPOP in project drill by apache.

the class TestNullInputMiniPlan method testRightMergeJoinEmptyBoth.

@Test
public void testRightMergeJoinEmptyBoth() throws Exception {
    final PhysicalOperator join = new MergeJoinPOP(null, null, Lists.newArrayList(joinCond("a", "EQUALS", "b")), JoinRelType.RIGHT);
    testTwoInputNullBatchHandling(join);
}

Example 15 with MergeJoinPOP

use of org.apache.drill.exec.physical.config.MergeJoinPOP in project drill by apache.

the class TestOutputBatchSize method testMergeJoinUpperLimit.

@Test
public void testMergeJoinUpperLimit() throws Exception {
    // test the upper limit of 65535 records per batch.
    MergeJoinPOP mergeJoin = new MergeJoinPOP(null, null, Lists.newArrayList(joinCond("c1", "EQUALS", "c2")), JoinRelType.LEFT);
    mockOpContext(mergeJoin, initReservation, maxAllocation);
    numRows = 100000;
    // create left input rows like this.
    // "a1" : 5,  "c1" : <id>
    List<String> leftJsonBatches = Lists.newArrayList();
    StringBuilder leftBatchString = new StringBuilder();
    leftBatchString.append("[");
    for (int i = 0; i < numRows; i++) {
        leftBatchString.append("{\"a1\": 5, " + "\"c1\" : " + i + "},");
    }
    leftBatchString.append("{\"a1\": 5, " + "\"c1\" : " + numRows + "}");
    leftBatchString.append("]");
    leftJsonBatches.add(leftBatchString.toString());
    // create right input rows like this.
    // "a2" : 6, "c2" : <id>
    List<String> rightJsonBatches = Lists.newArrayList();
    StringBuilder rightBatchString = new StringBuilder();
    rightBatchString.append("[");
    for (int i = 0; i < numRows; i++) {
        rightBatchString.append("{\"a2\": 6, " + "\"c2\" : " + i + "},");
    }
    rightBatchString.append("{\"a2\": 6, " + "\"c2\" : " + numRows + "}");
    rightBatchString.append("]");
    rightJsonBatches.add(rightBatchString.toString());
    // output rows will be like this.
    // "a1" : 5,  "c1" : 1, "a2":6,  "c2": 1
    // "a1" : 5,  "c1" : 2, "a2":6,  "c2": 2
    // "a1" : 5,  "c1" : 3, "a2":6,  "c2": 3
    // expect two batches, batch limited by 65535 records
    LegacyOperatorTestBuilder opTestBuilder = legacyOpTestBuilder().physicalOperator(mergeJoin).baselineColumns("a1", "c1", "a2", "c2").expectedNumBatches(// verify number of batches
    2).inputDataStreamsJson(Lists.newArrayList(leftJsonBatches, rightJsonBatches));
    for (long i = 0; i < numRows + 1; i++) {
        opTestBuilder.baselineValues(5l, i, 6l, i);
    }
    opTestBuilder.go();
}