Search in sources :

Example 6 with LegacyOperatorTestBuilder

use of org.apache.drill.test.LegacyOperatorTestBuilder in project drill by apache.

the class TestOutputBatchSize method testFlattenUpperLimit.

@Test
public void testFlattenUpperLimit() throws Exception {
    // test the upper limit of 65535 records per batch.
    PhysicalOperator flatten = new FlattenPOP(null, SchemaPath.getSimplePath("c"));
    mockOpContext(flatten, initReservation, maxAllocation);
    List<String> inputJsonBatches = Lists.newArrayList();
    StringBuilder batchString = new StringBuilder();
    StringBuilder flattenElement = new StringBuilder();
    // Create list of 1000 elements
    flattenElement.append("[");
    for (int i = 0; i < 1000; i++) {
        flattenElement.append(i);
        flattenElement.append(",");
    }
    flattenElement.append(1000);
    flattenElement.append("]");
    batchString.append("[");
    numRows = 100;
    for (int i = 0; i < numRows; i++) {
        batchString.append("{\"a\": 5, " + "\"c\":" + flattenElement + "},");
    }
    batchString.append("{\"a\": 5, " + "\"c\":" + flattenElement + "}");
    batchString.append("]");
    inputJsonBatches.add(batchString.toString());
    // Figure out what will be approximate total output size out of flatten for input above
    // We will use this sizing information to set output batch size so we can produce desired
    // number of batches that can be verified.
    List<String> expectedJsonBatches = Lists.newArrayList();
    StringBuilder expectedBatchString = new StringBuilder();
    expectedBatchString.append("[");
    for (int i = 0; i < numRows; i++) {
        for (int j = 0; j < 1000; j++) {
            expectedBatchString.append("{\"a\": 5, " + "\"c\" :");
            expectedBatchString.append(j);
            expectedBatchString.append("},");
        }
    }
    for (int j = 0; j < 999; j++) {
        expectedBatchString.append("{\"a\": 5, " + "\"c\" :");
        expectedBatchString.append(j);
        expectedBatchString.append("},");
    }
    expectedBatchString.append("{\"a\": 5, " + "\"c\" :");
    expectedBatchString.append(1000);
    expectedBatchString.append("}");
    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 16 batches because of upper bound of 65535 rows.
    fragContext.getOptions().setLocalOption("drill.exec.memory.operator.output_batch_size", totalSize / 2);
    // Here we expect 16 batches because each batch will be limited by upper limit of 65535 records.
    LegacyOperatorTestBuilder opTestBuilder = legacyOpTestBuilder().physicalOperator(flatten).inputDataStreamJson(inputJsonBatches).baselineColumns("a", "c").expectedNumBatches(// verify number of batches
    2).expectedBatchSize(// verify batch size.
    totalSize / 2);
    for (long i = 0; i < numRows + 1; i++) {
        for (long j = 0; j < 1001; j++) {
            opTestBuilder.baselineValues(5l, j);
        }
    }
    opTestBuilder.go();
}
Also used : FlattenPOP(org.apache.drill.exec.physical.config.FlattenPOP) LegacyOperatorTestBuilder(org.apache.drill.test.LegacyOperatorTestBuilder) PhysicalOperator(org.apache.drill.exec.physical.base.PhysicalOperator) Test(org.junit.Test)

Example 7 with LegacyOperatorTestBuilder

use of org.apache.drill.test.LegacyOperatorTestBuilder in project drill by apache.

the class TestOutputBatchSize method testLeftNestedLoopJoin.

@Test
public void testLeftNestedLoopJoin() throws Exception {
    LogicalExpression functionCallExpr = new FunctionCall("equal", ImmutableList.of((LogicalExpression) new FieldReference("c1", ExpressionPosition.UNKNOWN), (LogicalExpression) new FieldReference("c2", ExpressionPosition.UNKNOWN)), ExpressionPosition.UNKNOWN);
    NestedLoopJoinPOP nestedLoopJoin = new NestedLoopJoinPOP(null, null, JoinRelType.LEFT, functionCallExpr);
    numRows = 4000 * 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 1/2 of total size expected.
    // We will get approximately 4 batches.
    fragContext.getOptions().setLocalOption("drill.exec.memory.operator.output_batch_size", totalSize / 2);
    LegacyOperatorTestBuilder opTestBuilder = legacyOpTestBuilder().physicalOperator(nestedLoopJoin).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();
}
Also used : LogicalExpression(org.apache.drill.common.expression.LogicalExpression) FieldReference(org.apache.drill.common.expression.FieldReference) LegacyOperatorTestBuilder(org.apache.drill.test.LegacyOperatorTestBuilder) FunctionCall(org.apache.drill.common.expression.FunctionCall) NestedLoopJoinPOP(org.apache.drill.exec.physical.config.NestedLoopJoinPOP) Test(org.junit.Test)

Example 8 with LegacyOperatorTestBuilder

use of org.apache.drill.test.LegacyOperatorTestBuilder in project drill by apache.

the class TestOutputBatchSize method testUnionMultipleOutputBatches.

@Test
public void testUnionMultipleOutputBatches() throws Exception {
    UnionAll unionAll = new UnionAll(Collections.<PhysicalOperator>emptyList());
    mockOpContext(unionAll, initReservation, maxAllocation);
    // create multiple batches from both sides.
    numRows = 8000;
    // 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.
    // "a1" : 5, "b1" : wideString, "c1" : <id>
    List<String> rightJsonBatches = Lists.newArrayList();
    StringBuilder rightBatchString = new StringBuilder();
    rightBatchString.append("[");
    for (int i = 0; i < numRows; i++) {
        rightBatchString.append("{\"a1\": 5, " + "\"b1\" : " + "\"" + wideString + "\"," + "\"c1\" : " + i + "},");
    }
    rightBatchString.append("{\"a1\": 5, " + "\"b1\" : " + "\"" + wideString + "\"," + "\"c1\" : " + numRows + "}");
    rightBatchString.append("]");
    rightJsonBatches.add(rightBatchString.toString());
    // output rows will be like this.
    // "a1" : 5, "b1" : wideString, "c1" : 1
    // "a1" : 5, "b1" : wideString, "c1" : 2
    // "a1" : 5, "b1" : wideString, "c1" : 3
    List<String> expectedJsonBatches = Lists.newArrayList();
    StringBuilder expectedBatchString = new StringBuilder();
    expectedBatchString.append("[");
    for (int i = 0; i < numRows * 2; i++) {
        expectedBatchString.append("{\"a1\": 5, " + "\"b1\" : " + "\"" + wideString + "\"," + "\"c1\" : " + i + "},");
    }
    expectedBatchString.append("{\"a1\": 5, " + "\"b1\" : " + "\"" + wideString + "\"," + "\"c1\" : " + numRows + "}");
    expectedBatchString.append("]");
    expectedJsonBatches.add(expectedBatchString.toString());
    long totalSize = getExpectedSize(expectedJsonBatches);
    // set the output batch size to half of total size expected.
    // We should get 4 batches, 2 for the left and 2 for the right.
    fragContext.getOptions().setLocalOption("drill.exec.memory.operator.output_batch_size", totalSize / 2);
    LegacyOperatorTestBuilder opTestBuilder = legacyOpTestBuilder().physicalOperator(unionAll).baselineColumns("a1", "b1", "c1").expectedNumBatches(// verify number of batches
    4).expectedBatchSize(// verify batch size
    totalSize).inputDataStreamsJson(Lists.newArrayList(leftJsonBatches, rightJsonBatches));
    for (long i = 0; i < numRows + 1; i++) {
        opTestBuilder.baselineValues(5l, wideString, i);
    }
    for (long i = 0; i < numRows + 1; i++) {
        opTestBuilder.baselineValues(5l, wideString, i);
    }
    opTestBuilder.go();
}
Also used : LegacyOperatorTestBuilder(org.apache.drill.test.LegacyOperatorTestBuilder) UnionAll(org.apache.drill.exec.physical.config.UnionAll) Test(org.junit.Test)

Example 9 with LegacyOperatorTestBuilder

use of org.apache.drill.test.LegacyOperatorTestBuilder in project drill by apache.

the class TestOutputBatchSize method testNestedLoopJoinMultipleOutputBatches.

@Test
public void testNestedLoopJoinMultipleOutputBatches() throws Exception {
    LogicalExpression functionCallExpr = new FunctionCall("equal", ImmutableList.of((LogicalExpression) new FieldReference("c1", ExpressionPosition.UNKNOWN), (LogicalExpression) new FieldReference("c2", ExpressionPosition.UNKNOWN)), ExpressionPosition.UNKNOWN);
    NestedLoopJoinPOP nestedLoopJoin = new NestedLoopJoinPOP(null, null, JoinRelType.INNER, functionCallExpr);
    mockOpContext(nestedLoopJoin, initReservation, maxAllocation);
    numRows = 4000 * 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 1/2 of total size expected.
    // We will get approximately 4 batches.
    fragContext.getOptions().setLocalOption("drill.exec.memory.operator.output_batch_size", totalSize / 2);
    LegacyOperatorTestBuilder opTestBuilder = legacyOpTestBuilder().physicalOperator(nestedLoopJoin).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();
}
Also used : LogicalExpression(org.apache.drill.common.expression.LogicalExpression) FieldReference(org.apache.drill.common.expression.FieldReference) LegacyOperatorTestBuilder(org.apache.drill.test.LegacyOperatorTestBuilder) FunctionCall(org.apache.drill.common.expression.FunctionCall) NestedLoopJoinPOP(org.apache.drill.exec.physical.config.NestedLoopJoinPOP) Test(org.junit.Test)

Example 10 with LegacyOperatorTestBuilder

use of org.apache.drill.test.LegacyOperatorTestBuilder in project drill by apache.

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);
    LegacyOperatorTestBuilder opTestBuilder = legacyOpTestBuilder().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();
}
Also used : MergeJoinPOP(org.apache.drill.exec.physical.config.MergeJoinPOP) LegacyOperatorTestBuilder(org.apache.drill.test.LegacyOperatorTestBuilder) Test(org.junit.Test)

Aggregations

LegacyOperatorTestBuilder (org.apache.drill.test.LegacyOperatorTestBuilder)39 Test (org.junit.Test)36 PhysicalOperator (org.apache.drill.exec.physical.base.PhysicalOperator)12 FlattenPOP (org.apache.drill.exec.physical.config.FlattenPOP)12 HashJoinPOP (org.apache.drill.exec.physical.config.HashJoinPOP)6 FieldReference (org.apache.drill.common.expression.FieldReference)5 FunctionCall (org.apache.drill.common.expression.FunctionCall)5 LogicalExpression (org.apache.drill.common.expression.LogicalExpression)5 NestedLoopJoinPOP (org.apache.drill.exec.physical.config.NestedLoopJoinPOP)5 Project (org.apache.drill.exec.physical.config.Project)5 Text (org.apache.drill.exec.util.Text)5 MergeJoinPOP (org.apache.drill.exec.physical.config.MergeJoinPOP)4 HashAggregate (org.apache.drill.exec.physical.config.HashAggregate)3 UnionAll (org.apache.drill.exec.physical.config.UnionAll)3 JsonStringArrayList (org.apache.drill.exec.util.JsonStringArrayList)3 JsonStringHashMap (org.apache.drill.exec.util.JsonStringHashMap)3 MinorFragmentEndpoint (org.apache.drill.exec.physical.MinorFragmentEndpoint)2 ExternalSort (org.apache.drill.exec.physical.config.ExternalSort)1