Search in sources :

Example 1 with PerSegmentOptimizingQueryRunner

use of org.apache.druid.query.PerSegmentOptimizingQueryRunner in project druid by druid-io.

the class ServerManager method buildAndDecorateQueryRunner.

private <T> QueryRunner<T> buildAndDecorateQueryRunner(final QueryRunnerFactory<T, Query<T>> factory, final QueryToolChest<T, Query<T>> toolChest, final SegmentReference segment, final Optional<byte[]> cacheKeyPrefix, final SegmentDescriptor segmentDescriptor, final AtomicLong cpuTimeAccumulator) {
    final SpecificSegmentSpec segmentSpec = new SpecificSegmentSpec(segmentDescriptor);
    final SegmentId segmentId = segment.getId();
    final Interval segmentInterval = segment.getDataInterval();
    // If the segment is closed after this line, ReferenceCountingSegmentQueryRunner will handle and do the right thing.
    if (segmentId == null || segmentInterval == null) {
        return new ReportTimelineMissingSegmentQueryRunner<>(segmentDescriptor);
    }
    String segmentIdString = segmentId.toString();
    MetricsEmittingQueryRunner<T> metricsEmittingQueryRunnerInner = new MetricsEmittingQueryRunner<>(emitter, toolChest, new ReferenceCountingSegmentQueryRunner<>(factory, segment, segmentDescriptor), QueryMetrics::reportSegmentTime, queryMetrics -> queryMetrics.segment(segmentIdString));
    StorageAdapter storageAdapter = segment.asStorageAdapter();
    long segmentMaxTime = storageAdapter.getMaxTime().getMillis();
    long segmentMinTime = storageAdapter.getMinTime().getMillis();
    Interval actualDataInterval = Intervals.utc(segmentMinTime, segmentMaxTime + 1);
    CachingQueryRunner<T> cachingQueryRunner = new CachingQueryRunner<>(segmentIdString, cacheKeyPrefix, segmentDescriptor, actualDataInterval, objectMapper, cache, toolChest, metricsEmittingQueryRunnerInner, cachePopulator, cacheConfig);
    BySegmentQueryRunner<T> bySegmentQueryRunner = new BySegmentQueryRunner<>(segmentId, segmentInterval.getStart(), cachingQueryRunner);
    MetricsEmittingQueryRunner<T> metricsEmittingQueryRunnerOuter = new MetricsEmittingQueryRunner<>(emitter, toolChest, bySegmentQueryRunner, QueryMetrics::reportSegmentAndCacheTime, queryMetrics -> queryMetrics.segment(segmentIdString)).withWaitMeasuredFromNow();
    SpecificSegmentQueryRunner<T> specificSegmentQueryRunner = new SpecificSegmentQueryRunner<>(metricsEmittingQueryRunnerOuter, segmentSpec);
    PerSegmentOptimizingQueryRunner<T> perSegmentOptimizingQueryRunner = new PerSegmentOptimizingQueryRunner<>(specificSegmentQueryRunner, new PerSegmentQueryOptimizationContext(segmentDescriptor));
    return new SetAndVerifyContextQueryRunner<>(serverConfig, CPUTimeMetricQueryRunner.safeBuild(perSegmentOptimizingQueryRunner, toolChest, emitter, cpuTimeAccumulator, false));
}
Also used : SegmentManager(org.apache.druid.server.SegmentManager) Inject(com.google.inject.Inject) Smile(org.apache.druid.guice.annotations.Smile) QueryProcessingPool(org.apache.druid.query.QueryProcessingPool) StorageAdapter(org.apache.druid.segment.StorageAdapter) NoopQueryRunner(org.apache.druid.query.NoopQueryRunner) SegmentReference(org.apache.druid.segment.SegmentReference) SpecificSegmentQueryRunner(org.apache.druid.query.spec.SpecificSegmentQueryRunner) QueryRunner(org.apache.druid.query.QueryRunner) FinalizeResultsQueryRunner(org.apache.druid.query.FinalizeResultsQueryRunner) ReportTimelineMissingSegmentQueryRunner(org.apache.druid.query.ReportTimelineMissingSegmentQueryRunner) CacheConfig(org.apache.druid.client.cache.CacheConfig) StringUtils(org.apache.druid.java.util.common.StringUtils) JoinableFactoryWrapper(org.apache.druid.segment.join.JoinableFactoryWrapper) ISE(org.apache.druid.java.util.common.ISE) SpecificSegmentSpec(org.apache.druid.query.spec.SpecificSegmentSpec) BySegmentQueryRunner(org.apache.druid.query.BySegmentQueryRunner) SetAndVerifyContextQueryRunner(org.apache.druid.server.SetAndVerifyContextQueryRunner) QueryDataSource(org.apache.druid.query.QueryDataSource) PerSegmentQueryOptimizationContext(org.apache.druid.query.PerSegmentQueryOptimizationContext) ServiceEmitter(org.apache.druid.java.util.emitter.service.ServiceEmitter) Optional(java.util.Optional) PerSegmentOptimizingQueryRunner(org.apache.druid.query.PerSegmentOptimizingQueryRunner) FunctionalIterable(org.apache.druid.java.util.common.guava.FunctionalIterable) SegmentId(org.apache.druid.timeline.SegmentId) DataSourceAnalysis(org.apache.druid.query.planning.DataSourceAnalysis) Intervals(org.apache.druid.java.util.common.Intervals) QueryMetrics(org.apache.druid.query.QueryMetrics) CachingQueryRunner(org.apache.druid.client.CachingQueryRunner) JoinableFactory(org.apache.druid.segment.join.JoinableFactory) Function(java.util.function.Function) PartitionChunk(org.apache.druid.timeline.partition.PartitionChunk) Interval(org.joda.time.Interval) Lists(com.google.common.collect.Lists) MetricsEmittingQueryRunner(org.apache.druid.query.MetricsEmittingQueryRunner) Query(org.apache.druid.query.Query) CachePopulator(org.apache.druid.client.cache.CachePopulator) QuerySegmentWalker(org.apache.druid.query.QuerySegmentWalker) EmittingLogger(org.apache.druid.java.util.emitter.EmittingLogger) VersionedIntervalTimeline(org.apache.druid.timeline.VersionedIntervalTimeline) ServerConfig(org.apache.druid.server.initialization.ServerConfig) ObjectMapper(com.fasterxml.jackson.databind.ObjectMapper) QueryRunnerFactoryConglomerate(org.apache.druid.query.QueryRunnerFactoryConglomerate) QueryToolChest(org.apache.druid.query.QueryToolChest) ReferenceCountingSegment(org.apache.druid.segment.ReferenceCountingSegment) AtomicLong(java.util.concurrent.atomic.AtomicLong) ReferenceCountingSegmentQueryRunner(org.apache.druid.query.ReferenceCountingSegmentQueryRunner) QueryRunnerFactory(org.apache.druid.query.QueryRunnerFactory) SegmentDescriptor(org.apache.druid.query.SegmentDescriptor) Cache(org.apache.druid.client.cache.Cache) Filters(org.apache.druid.segment.filter.Filters) Collections(java.util.Collections) CPUTimeMetricQueryRunner(org.apache.druid.query.CPUTimeMetricQueryRunner) QueryUnsupportedException(org.apache.druid.query.QueryUnsupportedException) SegmentId(org.apache.druid.timeline.SegmentId) StorageAdapter(org.apache.druid.segment.StorageAdapter) BySegmentQueryRunner(org.apache.druid.query.BySegmentQueryRunner) MetricsEmittingQueryRunner(org.apache.druid.query.MetricsEmittingQueryRunner) PerSegmentQueryOptimizationContext(org.apache.druid.query.PerSegmentQueryOptimizationContext) ReportTimelineMissingSegmentQueryRunner(org.apache.druid.query.ReportTimelineMissingSegmentQueryRunner) SpecificSegmentSpec(org.apache.druid.query.spec.SpecificSegmentSpec) SpecificSegmentQueryRunner(org.apache.druid.query.spec.SpecificSegmentQueryRunner) CachingQueryRunner(org.apache.druid.client.CachingQueryRunner) SetAndVerifyContextQueryRunner(org.apache.druid.server.SetAndVerifyContextQueryRunner) QueryMetrics(org.apache.druid.query.QueryMetrics) PerSegmentOptimizingQueryRunner(org.apache.druid.query.PerSegmentOptimizingQueryRunner) Interval(org.joda.time.Interval)

Example 2 with PerSegmentOptimizingQueryRunner

use of org.apache.druid.query.PerSegmentOptimizingQueryRunner in project druid by druid-io.

the class TimeCompareBenchmark method setup.

@Setup
public void setup() throws IOException {
    log.info("SETUP CALLED AT " + System.currentTimeMillis());
    ComplexMetrics.registerSerde("hyperUnique", new HyperUniquesSerde());
    executorService = Execs.multiThreaded(numSegments, "TopNThreadPool");
    setupQueries();
    String schemaName = "basic";
    schemaInfo = GeneratorBasicSchemas.SCHEMA_MAP.get(schemaName);
    segmentIntervals = new Interval[numSegments];
    long startMillis = schemaInfo.getDataInterval().getStartMillis();
    long endMillis = schemaInfo.getDataInterval().getEndMillis();
    long partialIntervalMillis = (endMillis - startMillis) / numSegments;
    for (int i = 0; i < numSegments; i++) {
        long partialEndMillis = startMillis + partialIntervalMillis;
        segmentIntervals[i] = Intervals.utc(startMillis, partialEndMillis);
        log.info("Segment [%d] with interval [%s]", i, segmentIntervals[i]);
        startMillis = partialEndMillis;
    }
    incIndexes = new ArrayList<>();
    for (int i = 0; i < numSegments; i++) {
        log.info("Generating rows for segment " + i);
        DataGenerator gen = new DataGenerator(schemaInfo.getColumnSchemas(), RNG_SEED + i, segmentIntervals[i], 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);
        }
        incIndexes.add(incIndex);
    }
    tmpDir = FileUtils.createTempDir();
    log.info("Using temp dir: " + tmpDir.getAbsolutePath());
    qIndexes = new ArrayList<>();
    for (int i = 0; i < numSegments; i++) {
        File indexFile = INDEX_MERGER_V9.persist(incIndexes.get(i), tmpDir, new IndexSpec(), null);
        QueryableIndex qIndex = INDEX_IO.loadIndex(indexFile);
        qIndexes.add(qIndex);
    }
    List<QueryRunner<Result<TopNResultValue>>> singleSegmentRunners = new ArrayList<>();
    QueryToolChest toolChest = topNFactory.getToolchest();
    for (int i = 0; i < numSegments; i++) {
        SegmentId segmentId = SegmentId.dummy("qIndex " + i);
        QueryRunner<Result<TopNResultValue>> runner = QueryBenchmarkUtil.makeQueryRunner(topNFactory, segmentId, new QueryableIndexSegment(qIndexes.get(i), segmentId));
        singleSegmentRunners.add(new PerSegmentOptimizingQueryRunner<>(toolChest.preMergeQueryDecoration(runner), new PerSegmentQueryOptimizationContext(new SegmentDescriptor(segmentIntervals[i], "1", 0))));
    }
    topNRunner = toolChest.postMergeQueryDecoration(new FinalizeResultsQueryRunner<>(toolChest.mergeResults(topNFactory.mergeRunners(executorService, singleSegmentRunners)), toolChest));
    List<QueryRunner<Result<TimeseriesResultValue>>> singleSegmentRunnersT = new ArrayList<>();
    QueryToolChest toolChestT = timeseriesFactory.getToolchest();
    for (int i = 0; i < numSegments; i++) {
        SegmentId segmentId = SegmentId.dummy("qIndex " + i);
        QueryRunner<Result<TimeseriesResultValue>> runner = QueryBenchmarkUtil.makeQueryRunner(timeseriesFactory, segmentId, new QueryableIndexSegment(qIndexes.get(i), segmentId));
        singleSegmentRunnersT.add(new PerSegmentOptimizingQueryRunner<>(toolChestT.preMergeQueryDecoration(runner), new PerSegmentQueryOptimizationContext(new SegmentDescriptor(segmentIntervals[i], "1", 0))));
    }
    timeseriesRunner = toolChestT.postMergeQueryDecoration(new FinalizeResultsQueryRunner<>(toolChestT.mergeResults(timeseriesFactory.mergeRunners(executorService, singleSegmentRunnersT)), toolChestT));
}
Also used : QueryableIndexSegment(org.apache.druid.segment.QueryableIndexSegment) TimeseriesResultValue(org.apache.druid.query.timeseries.TimeseriesResultValue) IndexSpec(org.apache.druid.segment.IndexSpec) ArrayList(java.util.ArrayList) HyperUniquesSerde(org.apache.druid.query.aggregation.hyperloglog.HyperUniquesSerde) TopNQueryQueryToolChest(org.apache.druid.query.topn.TopNQueryQueryToolChest) TimeseriesQueryQueryToolChest(org.apache.druid.query.timeseries.TimeseriesQueryQueryToolChest) QueryToolChest(org.apache.druid.query.QueryToolChest) Result(org.apache.druid.query.Result) SegmentDescriptor(org.apache.druid.query.SegmentDescriptor) TopNResultValue(org.apache.druid.query.topn.TopNResultValue) IncrementalIndex(org.apache.druid.segment.incremental.IncrementalIndex) OnheapIncrementalIndex(org.apache.druid.segment.incremental.OnheapIncrementalIndex) SegmentId(org.apache.druid.timeline.SegmentId) QueryRunner(org.apache.druid.query.QueryRunner) FinalizeResultsQueryRunner(org.apache.druid.query.FinalizeResultsQueryRunner) PerSegmentOptimizingQueryRunner(org.apache.druid.query.PerSegmentOptimizingQueryRunner) PerSegmentQueryOptimizationContext(org.apache.druid.query.PerSegmentQueryOptimizationContext) FinalizeResultsQueryRunner(org.apache.druid.query.FinalizeResultsQueryRunner) QueryableIndex(org.apache.druid.segment.QueryableIndex) DataGenerator(org.apache.druid.segment.generator.DataGenerator) InputRow(org.apache.druid.data.input.InputRow) File(java.io.File) Setup(org.openjdk.jmh.annotations.Setup)

Aggregations

FinalizeResultsQueryRunner (org.apache.druid.query.FinalizeResultsQueryRunner)2 PerSegmentOptimizingQueryRunner (org.apache.druid.query.PerSegmentOptimizingQueryRunner)2 PerSegmentQueryOptimizationContext (org.apache.druid.query.PerSegmentQueryOptimizationContext)2 QueryRunner (org.apache.druid.query.QueryRunner)2 QueryToolChest (org.apache.druid.query.QueryToolChest)2 SegmentDescriptor (org.apache.druid.query.SegmentDescriptor)2 ObjectMapper (com.fasterxml.jackson.databind.ObjectMapper)1 Lists (com.google.common.collect.Lists)1 Inject (com.google.inject.Inject)1 File (java.io.File)1 ArrayList (java.util.ArrayList)1 Collections (java.util.Collections)1 Optional (java.util.Optional)1 AtomicLong (java.util.concurrent.atomic.AtomicLong)1 Function (java.util.function.Function)1 CachingQueryRunner (org.apache.druid.client.CachingQueryRunner)1 Cache (org.apache.druid.client.cache.Cache)1 CacheConfig (org.apache.druid.client.cache.CacheConfig)1 CachePopulator (org.apache.druid.client.cache.CachePopulator)1 InputRow (org.apache.druid.data.input.InputRow)1