private void testQueryGeneric(
long orderSize,
long lineItemSize,
float ordersFilterFactor,
float joinFilterFactor,
boolean broadcastOkay,
boolean partitionedOkay,
boolean hashJoinFirstOkay,
boolean hashJoinSecondOkay,
boolean mergeJoinOkay) {
TPCHQuery3 query = new TPCHQuery3();
Plan p = query.getPlan(DEFAULT_PARALLELISM_STRING, IN_FILE, IN_FILE, OUT_FILE);
p.setExecutionConfig(defaultExecutionConfig);
testQueryGeneric(
p,
orderSize,
lineItemSize,
ordersFilterFactor,
joinFilterFactor,
broadcastOkay,
partitionedOkay,
hashJoinFirstOkay,
hashJoinSecondOkay,
mergeJoinOkay);
}
/**
* Verifies that a robust repartitioning plan with a hash join is created in the absence of
* statistics.
*/
@Test
public void testQueryNoStatistics() {
try {
TPCHQuery3 query = new TPCHQuery3();
Plan p = query.getPlan(DEFAULT_PARALLELISM_STRING, IN_FILE, IN_FILE, OUT_FILE);
p.setExecutionConfig(defaultExecutionConfig);
// compile
final OptimizedPlan plan = compileNoStats(p);
final OptimizerPlanNodeResolver or = getOptimizerPlanNodeResolver(plan);
// get the nodes from the final plan
final SinkPlanNode sink = or.getNode("Output");
final SingleInputPlanNode reducer = or.getNode("AggLio");
final SingleInputPlanNode combiner =
reducer.getPredecessor() instanceof SingleInputPlanNode
? (SingleInputPlanNode) reducer.getPredecessor()
: null;
final DualInputPlanNode join = or.getNode("JoinLiO");
final SingleInputPlanNode filteringMapper = or.getNode("FilterO");
// verify the optimizer choices
checkStandardStrategies(filteringMapper, join, combiner, reducer, sink);
Assert.assertTrue(checkRepartitionShipStrategies(join, reducer, combiner));
Assert.assertTrue(
checkHashJoinStrategies(join, reducer, true)
|| checkHashJoinStrategies(join, reducer, false));
} catch (Exception e) {
e.printStackTrace();
Assert.fail(e.getMessage());
}
}
/**
* Statistics that push towards a repartition merge join. If the join blows the data volume up
* significantly, re-exploiting the sorted order is cheaper.
*/
@Test
public void testQueryWithStatsForRepartitionMerge() {
TPCHQuery3 query = new TPCHQuery3();
Plan p = query.getPlan(DEFAULT_PARALLELISM_STRING, IN_FILE, IN_FILE, OUT_FILE);
p.setExecutionConfig(defaultExecutionConfig);
// set compiler hints
OperatorResolver cr = getContractResolver(p);
JoinOperator match = cr.getNode("JoinLiO");
match.getCompilerHints().setFilterFactor(100f);
testQueryGeneric(
100l * 1024 * 1024 * 1024 * 1024,
100l * 1024 * 1024 * 1024 * 1024,
0.05f,
100f,
false,
true,
false,
false,
true);
}