|
| 1 | +// Copyright (c) Meta Platforms, Inc. and affiliates. |
| 2 | +// This source code is licensed under both the GPLv2 (found in the |
| 3 | +// COPYING file in the root directory) and Apache 2.0 License |
| 4 | +// (found in the LICENSE.Apache file in the root directory). |
| 5 | +// |
| 6 | + |
| 7 | +#include "util/auto_skip_compressor.h" |
| 8 | + |
| 9 | +#include "options/options_helper.h" |
| 10 | +#include "rocksdb/advanced_compression.h" |
| 11 | +#include "util/random.h" |
| 12 | +namespace ROCKSDB_NAMESPACE { |
| 13 | + |
| 14 | +int CompressionRejectionProbabilityPredictor::Predict() const { |
| 15 | + return pred_rejection_prob_percentage_; |
| 16 | +} |
| 17 | + |
| 18 | +size_t CompressionRejectionProbabilityPredictor::attempted_compression_count() |
| 19 | + const { |
| 20 | + return rejected_count_ + compressed_count_; |
| 21 | +} |
| 22 | + |
| 23 | +bool CompressionRejectionProbabilityPredictor::Record( |
| 24 | + Slice uncompressed_block_data, std::string* compressed_output, |
| 25 | + const CompressionOptions& opts) { |
| 26 | + if (compressed_output->size() > |
| 27 | + (static_cast<uint64_t>(opts.max_compressed_bytes_per_kb) * |
| 28 | + uncompressed_block_data.size()) >> |
| 29 | + 10) { |
| 30 | + rejected_count_++; |
| 31 | + } else { |
| 32 | + compressed_count_++; |
| 33 | + } |
| 34 | + if (attempted_compression_count() >= window_size_) { |
| 35 | + pred_rejection_prob_percentage_ = static_cast<int>( |
| 36 | + rejected_count_ * 100 / (compressed_count_ + rejected_count_)); |
| 37 | + compressed_count_ = 0; |
| 38 | + rejected_count_ = 0; |
| 39 | + assert(attempted_compression_count() == 0); |
| 40 | + } |
| 41 | + return true; |
| 42 | +} |
| 43 | +AutoSkipCompressorWrapper::AutoSkipCompressorWrapper( |
| 44 | + std::unique_ptr<Compressor> compressor, const CompressionOptions& opts, |
| 45 | + const CompressionType type) |
| 46 | + : CompressorWrapper::CompressorWrapper(std::move(compressor)), |
| 47 | + opts_(opts), |
| 48 | + type_(type), |
| 49 | + predictor_( |
| 50 | + std::make_shared<CompressionRejectionProbabilityPredictor>(10)) { |
| 51 | + (void)type_; |
| 52 | + (void)opts_; |
| 53 | +} |
| 54 | + |
| 55 | +Status AutoSkipCompressorWrapper::CompressBlock( |
| 56 | + Slice uncompressed_data, std::string* compressed_output, |
| 57 | + CompressionType* out_compression_type, ManagedWorkingArea* wa) { |
| 58 | + bool exploration = |
| 59 | + Random::GetTLSInstance()->PercentTrue(kExplorationPercentage); |
| 60 | + TEST_SYNC_POINT_CALLBACK( |
| 61 | + "AutoSkipCompressorWrapper::CompressBlock::exploitOrExplore", |
| 62 | + &exploration); |
| 63 | + if (exploration) { |
| 64 | + return CompressBlockAndRecord(uncompressed_data, compressed_output, |
| 65 | + out_compression_type, wa); |
| 66 | + } else { |
| 67 | + auto prediction = predictor_->Predict(); |
| 68 | + if (prediction <= kProbabilityCutOff) { |
| 69 | + // decide to compress |
| 70 | + return CompressBlockAndRecord(uncompressed_data, compressed_output, |
| 71 | + out_compression_type, wa); |
| 72 | + } else { |
| 73 | + // decide to bypass compression |
| 74 | + *out_compression_type = kNoCompression; |
| 75 | + return Status::OK(); |
| 76 | + } |
| 77 | + } |
| 78 | + return Status::OK(); |
| 79 | +} |
| 80 | + |
| 81 | +Status AutoSkipCompressorWrapper::CompressBlockAndRecord( |
| 82 | + Slice uncompressed_data, std::string* compressed_output, |
| 83 | + CompressionType* out_compression_type, ManagedWorkingArea* wa) { |
| 84 | + Status status = wrapped_->CompressBlock(uncompressed_data, compressed_output, |
| 85 | + out_compression_type, wa); |
| 86 | + // determine if it was rejected or compressed |
| 87 | + predictor_->Record(uncompressed_data, compressed_output, opts_); |
| 88 | + return status; |
| 89 | +} |
| 90 | + |
| 91 | +const char* AutoSkipCompressorManager::Name() const { |
| 92 | + // should have returned "AutoSkipCompressorManager" but we currently have an |
| 93 | + // error so for now returning name of the wrapped container |
| 94 | + return wrapped_->Name(); |
| 95 | +} |
| 96 | + |
| 97 | +std::unique_ptr<Compressor> AutoSkipCompressorManager::GetCompressorForSST( |
| 98 | + const FilterBuildingContext& context, const CompressionOptions& opts, |
| 99 | + CompressionType preferred) { |
| 100 | + assert(GetSupportedCompressions().size() > 1); |
| 101 | + assert(preferred != kNoCompression); |
| 102 | + return std::make_unique<AutoSkipCompressorWrapper>( |
| 103 | + wrapped_->GetCompressorForSST(context, opts, preferred), opts, preferred); |
| 104 | +} |
| 105 | + |
| 106 | +std::shared_ptr<CompressionManagerWrapper> CreateAutoSkipCompressionManager( |
| 107 | + std::shared_ptr<CompressionManager> wrapped) { |
| 108 | + return std::make_shared<AutoSkipCompressorManager>( |
| 109 | + wrapped == nullptr ? GetDefaultBuiltinCompressionManager() : wrapped); |
| 110 | +} |
| 111 | +} // namespace ROCKSDB_NAMESPACE |
0 commit comments