@Setup
public void setup() throws IOException {
log.info("SETUP CALLED AT " + System.currentTimeMillis());
if (ComplexMetrics.getSerdeForType("hyperUnique") == null) {
ComplexMetrics.registerSerde("hyperUnique", new HyperUniquesSerde(Hashing.murmur3_128()));
}
executorService = Execs.multiThreaded(numSegments, "TimeseriesThreadPool");
setupQueries();
String[] schemaQuery = schemaAndQuery.split("\\.");
String schemaName = schemaQuery[0];
String queryName = schemaQuery[1];
schemaInfo = BenchmarkSchemas.SCHEMA_MAP.get(schemaName);
query = SCHEMA_QUERY_MAP.get(schemaName).get(queryName);
incIndexes = new ArrayList<>();
for (int i = 0; i < numSegments; i++) {
log.info("Generating rows for segment " + i);
BenchmarkDataGenerator gen =
new BenchmarkDataGenerator(
schemaInfo.getColumnSchemas(),
RNG_SEED + i,
schemaInfo.getDataInterval(),
rowsPerSegment);
IncrementalIndex incIndex = makeIncIndex();
for (int j = 0; j < rowsPerSegment; j++) {
InputRow row = gen.nextRow();
if (j % 10000 == 0) {
log.info(j + " rows generated.");
}
incIndex.add(row);
}
log.info(rowsPerSegment + " rows generated");
incIndexes.add(incIndex);
}
File tmpFile = Files.createTempDir();
log.info("Using temp dir: " + tmpFile.getAbsolutePath());
tmpFile.deleteOnExit();
qIndexes = new ArrayList<>();
for (int i = 0; i < numSegments; i++) {
File indexFile = INDEX_MERGER_V9.persist(incIndexes.get(i), tmpFile, new IndexSpec());
QueryableIndex qIndex = INDEX_IO.loadIndex(indexFile);
qIndexes.add(qIndex);
}
factory =
new TimeseriesQueryRunnerFactory(
new TimeseriesQueryQueryToolChest(
QueryBenchmarkUtil.NoopIntervalChunkingQueryRunnerDecorator()),
new TimeseriesQueryEngine(),
QueryBenchmarkUtil.NOOP_QUERYWATCHER);
}
private void setupQueries() {
// queries for the basic schema
Map<String, TimeseriesQuery> basicQueries = new LinkedHashMap<>();
BenchmarkSchemaInfo basicSchema = BenchmarkSchemas.SCHEMA_MAP.get("basic");
{ // basic.A
QuerySegmentSpec intervalSpec =
new MultipleIntervalSegmentSpec(Arrays.asList(basicSchema.getDataInterval()));
List<AggregatorFactory> queryAggs = new ArrayList<>();
queryAggs.add(new LongSumAggregatorFactory("sumLongSequential", "sumLongSequential"));
queryAggs.add(new LongMaxAggregatorFactory("maxLongUniform", "maxLongUniform"));
queryAggs.add(new DoubleSumAggregatorFactory("sumFloatNormal", "sumFloatNormal"));
queryAggs.add(new DoubleMinAggregatorFactory("minFloatZipf", "minFloatZipf"));
queryAggs.add(new HyperUniquesAggregatorFactory("hyperUniquesMet", "hyper"));
TimeseriesQuery queryA =
Druids.newTimeseriesQueryBuilder()
.dataSource("blah")
.granularity(QueryGranularities.ALL)
.intervals(intervalSpec)
.aggregators(queryAggs)
.descending(false)
.build();
basicQueries.put("A", queryA);
}
{
QuerySegmentSpec intervalSpec =
new MultipleIntervalSegmentSpec(Arrays.asList(basicSchema.getDataInterval()));
List<AggregatorFactory> queryAggs = new ArrayList<>();
LongSumAggregatorFactory lsaf =
new LongSumAggregatorFactory("sumLongSequential", "sumLongSequential");
BoundDimFilter timeFilter =
new BoundDimFilter(
Column.TIME_COLUMN_NAME,
"200000",
"300000",
false,
false,
null,
null,
StringComparators.NUMERIC);
queryAggs.add(new FilteredAggregatorFactory(lsaf, timeFilter));
TimeseriesQuery timeFilterQuery =
Druids.newTimeseriesQueryBuilder()
.dataSource("blah")
.granularity(QueryGranularities.ALL)
.intervals(intervalSpec)
.aggregators(queryAggs)
.descending(false)
.build();
basicQueries.put("timeFilterNumeric", timeFilterQuery);
}
{
QuerySegmentSpec intervalSpec =
new MultipleIntervalSegmentSpec(Arrays.asList(basicSchema.getDataInterval()));
List<AggregatorFactory> queryAggs = new ArrayList<>();
LongSumAggregatorFactory lsaf =
new LongSumAggregatorFactory("sumLongSequential", "sumLongSequential");
BoundDimFilter timeFilter =
new BoundDimFilter(
Column.TIME_COLUMN_NAME,
"200000",
"300000",
false,
false,
null,
null,
StringComparators.ALPHANUMERIC);
queryAggs.add(new FilteredAggregatorFactory(lsaf, timeFilter));
TimeseriesQuery timeFilterQuery =
Druids.newTimeseriesQueryBuilder()
.dataSource("blah")
.granularity(QueryGranularities.ALL)
.intervals(intervalSpec)
.aggregators(queryAggs)
.descending(false)
.build();
basicQueries.put("timeFilterAlphanumeric", timeFilterQuery);
}
{
QuerySegmentSpec intervalSpec =
new MultipleIntervalSegmentSpec(Arrays.asList(new Interval(200000, 300000)));
List<AggregatorFactory> queryAggs = new ArrayList<>();
LongSumAggregatorFactory lsaf =
new LongSumAggregatorFactory("sumLongSequential", "sumLongSequential");
queryAggs.add(lsaf);
TimeseriesQuery timeFilterQuery =
Druids.newTimeseriesQueryBuilder()
.dataSource("blah")
.granularity(QueryGranularities.ALL)
.intervals(intervalSpec)
.aggregators(queryAggs)
.descending(false)
.build();
basicQueries.put("timeFilterByInterval", timeFilterQuery);
}
SCHEMA_QUERY_MAP.put("basic", basicQueries);
}