Examples with BloomFilterSpecification org.apache.hyracks.storage.am.bloomfilter.impls.BloomFilterSpecification used on opensource projects

Example 6 with BloomFilterSpecification

use of org.apache.hyracks.storage.am.bloomfilter.impls.BloomFilterSpecification in project asterixdb by apache.

the class ExternalRTree method merge.

// The only reason for overriding the merge method is the way to determine
// the need to keep deleted tuples
// This can be done in a better way by creating a method boolean
// keepDeletedTuples(mergedComponents);
@Override
public ILSMDiskComponent merge(ILSMIOOperation operation) throws HyracksDataException {
    LSMRTreeMergeOperation mergeOp = (LSMRTreeMergeOperation) operation;
    IIndexCursor cursor = mergeOp.getCursor();
    ISearchPredicate rtreeSearchPred = new SearchPredicate(null, null);
    ILSMIndexOperationContext opCtx = ((LSMRTreeSortedCursor) cursor).getOpCtx();
    opCtx.getComponentHolder().addAll(mergeOp.getMergingComponents());
    search(opCtx, cursor, rtreeSearchPred);
    LSMRTreeDiskComponent mergedComponent = createDiskComponent(componentFactory, mergeOp.getTarget(), mergeOp.getBTreeTarget(), mergeOp.getBloomFilterTarget(), true);
    // In case we must keep the deleted-keys BTrees, then they must be
    // merged *before* merging the r-trees so that
    // lsmHarness.endSearch() is called once when the r-trees have been
    // merged.
    boolean keepDeleteTuples = false;
    if (version == 0) {
        keepDeleteTuples = mergeOp.getMergingComponents().get(mergeOp.getMergingComponents().size() - 1) != diskComponents.get(diskComponents.size() - 1);
    } else {
        keepDeleteTuples = mergeOp.getMergingComponents().get(mergeOp.getMergingComponents().size() - 1) != secondDiskComponents.get(secondDiskComponents.size() - 1);
    }
    if (keepDeleteTuples) {
        // Keep the deleted tuples since the oldest disk component is not
        // included in the merge operation
        LSMRTreeDeletedKeysBTreeMergeCursor btreeCursor = new LSMRTreeDeletedKeysBTreeMergeCursor(opCtx);
        search(opCtx, btreeCursor, rtreeSearchPred);
        BTree btree = mergedComponent.getBTree();
        IIndexBulkLoader btreeBulkLoader = btree.createBulkLoader(1.0f, true, 0L, false);
        long numElements = 0L;
        for (int i = 0; i < mergeOp.getMergingComponents().size(); ++i) {
            numElements += ((LSMRTreeDiskComponent) mergeOp.getMergingComponents().get(i)).getBloomFilter().getNumElements();
        }
        int maxBucketsPerElement = BloomCalculations.maxBucketsPerElement(numElements);
        BloomFilterSpecification bloomFilterSpec = BloomCalculations.computeBloomSpec(maxBucketsPerElement, bloomFilterFalsePositiveRate);
        IIndexBulkLoader builder = mergedComponent.getBloomFilter().createBuilder(numElements, bloomFilterSpec.getNumHashes(), bloomFilterSpec.getNumBucketsPerElements());
        try {
            while (btreeCursor.hasNext()) {
                btreeCursor.next();
                ITupleReference tuple = btreeCursor.getTuple();
                btreeBulkLoader.add(tuple);
                builder.add(tuple);
            }
        } finally {
            btreeCursor.close();
            builder.end();
        }
        btreeBulkLoader.end();
    }
    IIndexBulkLoader bulkLoader = mergedComponent.getRTree().createBulkLoader(1.0f, false, 0L, false);
    try {
        while (cursor.hasNext()) {
            cursor.next();
            ITupleReference frameTuple = cursor.getTuple();
            bulkLoader.add(frameTuple);
        }
    } finally {
        cursor.close();
    }
    bulkLoader.end();
    return mergedComponent;
}

Example 7 with BloomFilterSpecification

use of org.apache.hyracks.storage.am.bloomfilter.impls.BloomFilterSpecification in project asterixdb by apache.

the class LSMInvertedIndex method createComponentBulkLoader.

@Override
public ILSMDiskComponentBulkLoader createComponentBulkLoader(ILSMDiskComponent component, float fillFactor, boolean verifyInput, long numElementsHint, boolean checkIfEmptyIndex, boolean withFilter) throws HyracksDataException {
    BloomFilterSpecification bloomFilterSpec = null;
    if (numElementsHint > 0) {
        int maxBucketsPerElement = BloomCalculations.maxBucketsPerElement(numElementsHint);
        bloomFilterSpec = BloomCalculations.computeBloomSpec(maxBucketsPerElement, bloomFilterFalsePositiveRate);
    }
    if (withFilter && filterFields != null) {
        return new LSMInvertedIndexDiskComponentBulkLoader((LSMInvertedIndexDiskComponent) component, bloomFilterSpec, fillFactor, verifyInput, numElementsHint, checkIfEmptyIndex, filterManager, treeFields, filterFields, MultiComparator.create(component.getLSMComponentFilter().getFilterCmpFactories()));
    } else {
        return new LSMInvertedIndexDiskComponentBulkLoader((LSMInvertedIndexDiskComponent) component, bloomFilterSpec, fillFactor, verifyInput, numElementsHint, checkIfEmptyIndex);
    }
}