De-template block based table iterator (#6531)
Summary:
Right now block based table iterator is used as both of iterating data for block based table, and for the index iterator for partitioend index. This was initially convenient for introducing a new iterator and block type for new index format, while reducing code change. However, these two usage doesn't go with each other very well. For example, Prev() is never called for partitioned index iterator, and some other complexity is maintained in block based iterators, which is not needed for index iterator but maintainers will always need to reason about it. Furthermore, the template usage is not following Google C++ Style which we are following, and makes a large chunk of code tangled together. This commit separate the two iterators. Right now, here is what it is done:
1. Copy the block based iterator code into partitioned index iterator, and de-template them.
2. Remove some code not needed for partitioned index. The upper bound check and tricks are removed. We never tested performance for those tricks when partitioned index is enabled in the first place. It's unlikelyl to generate performance regression, as creating new partitioned index block is much rarer than data blocks.
3. Separate out the prefetch logic to a helper class and both classes call them.
This commit will enable future follow-ups. One direction is that we might separate index iterator interface for data blocks and index blocks, as they are quite different.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6531
Test Plan: build using make and cmake. And build release
Differential Revision: D20473108
fbshipit-source-id: e48011783b339a4257c204cc07507b171b834b0f
2020-03-16 20:17:34 +01:00
|
|
|
// Copyright (c) 2011-present, Facebook, Inc. All rights reserved.
|
|
|
|
// This source code is licensed under both the GPLv2 (found in the
|
|
|
|
// COPYING file in the root directory) and Apache 2.0 License
|
|
|
|
// (found in the LICENSE.Apache file in the root directory).
|
|
|
|
//
|
|
|
|
// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
|
|
|
|
// Use of this source code is governed by a BSD-style license that can be
|
|
|
|
// found in the LICENSE file. See the AUTHORS file for names of contributors.
|
|
|
|
#pragma once
|
|
|
|
#include "table/block_based/block_based_table_reader.h"
|
|
|
|
|
|
|
|
namespace ROCKSDB_NAMESPACE {
|
|
|
|
class BlockPrefetcher {
|
|
|
|
public:
|
2022-04-16 02:28:09 +02:00
|
|
|
explicit BlockPrefetcher(size_t compaction_readahead_size,
|
|
|
|
size_t initial_auto_readahead_size)
|
|
|
|
: compaction_readahead_size_(compaction_readahead_size),
|
|
|
|
readahead_size_(initial_auto_readahead_size),
|
|
|
|
initial_auto_readahead_size_(initial_auto_readahead_size) {}
|
|
|
|
|
De-template block based table iterator (#6531)
Summary:
Right now block based table iterator is used as both of iterating data for block based table, and for the index iterator for partitioend index. This was initially convenient for introducing a new iterator and block type for new index format, while reducing code change. However, these two usage doesn't go with each other very well. For example, Prev() is never called for partitioned index iterator, and some other complexity is maintained in block based iterators, which is not needed for index iterator but maintainers will always need to reason about it. Furthermore, the template usage is not following Google C++ Style which we are following, and makes a large chunk of code tangled together. This commit separate the two iterators. Right now, here is what it is done:
1. Copy the block based iterator code into partitioned index iterator, and de-template them.
2. Remove some code not needed for partitioned index. The upper bound check and tricks are removed. We never tested performance for those tricks when partitioned index is enabled in the first place. It's unlikelyl to generate performance regression, as creating new partitioned index block is much rarer than data blocks.
3. Separate out the prefetch logic to a helper class and both classes call them.
This commit will enable future follow-ups. One direction is that we might separate index iterator interface for data blocks and index blocks, as they are quite different.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6531
Test Plan: build using make and cmake. And build release
Differential Revision: D20473108
fbshipit-source-id: e48011783b339a4257c204cc07507b171b834b0f
2020-03-16 20:17:34 +01:00
|
|
|
void PrefetchIfNeeded(const BlockBasedTable::Rep* rep,
|
|
|
|
const BlockHandle& handle, size_t readahead_size,
|
2022-03-21 15:12:43 +01:00
|
|
|
bool is_for_compaction, bool async_io);
|
De-template block based table iterator (#6531)
Summary:
Right now block based table iterator is used as both of iterating data for block based table, and for the index iterator for partitioend index. This was initially convenient for introducing a new iterator and block type for new index format, while reducing code change. However, these two usage doesn't go with each other very well. For example, Prev() is never called for partitioned index iterator, and some other complexity is maintained in block based iterators, which is not needed for index iterator but maintainers will always need to reason about it. Furthermore, the template usage is not following Google C++ Style which we are following, and makes a large chunk of code tangled together. This commit separate the two iterators. Right now, here is what it is done:
1. Copy the block based iterator code into partitioned index iterator, and de-template them.
2. Remove some code not needed for partitioned index. The upper bound check and tricks are removed. We never tested performance for those tricks when partitioned index is enabled in the first place. It's unlikelyl to generate performance regression, as creating new partitioned index block is much rarer than data blocks.
3. Separate out the prefetch logic to a helper class and both classes call them.
This commit will enable future follow-ups. One direction is that we might separate index iterator interface for data blocks and index blocks, as they are quite different.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6531
Test Plan: build using make and cmake. And build release
Differential Revision: D20473108
fbshipit-source-id: e48011783b339a4257c204cc07507b171b834b0f
2020-03-16 20:17:34 +01:00
|
|
|
FilePrefetchBuffer* prefetch_buffer() { return prefetch_buffer_.get(); }
|
|
|
|
|
2021-11-11 01:18:27 +01:00
|
|
|
void UpdateReadPattern(const uint64_t& offset, const size_t& len) {
|
2021-04-28 21:52:53 +02:00
|
|
|
prev_offset_ = offset;
|
|
|
|
prev_len_ = len;
|
|
|
|
}
|
|
|
|
|
2021-11-11 01:18:27 +01:00
|
|
|
bool IsBlockSequential(const uint64_t& offset) {
|
2021-04-28 21:52:53 +02:00
|
|
|
return (prev_len_ == 0 || (prev_offset_ + prev_len_ == offset));
|
|
|
|
}
|
|
|
|
|
2022-04-16 02:28:09 +02:00
|
|
|
void ResetValues(size_t initial_auto_readahead_size) {
|
2021-04-28 21:52:53 +02:00
|
|
|
num_file_reads_ = 1;
|
2021-12-01 07:52:14 +01:00
|
|
|
// Since initial_auto_readahead_size_ can be different from
|
2022-04-16 02:28:09 +02:00
|
|
|
// the value passed to BlockBasedTableOptions.initial_auto_readahead_size in
|
|
|
|
// case of adaptive_readahead, so fallback the readahead_size_ to that value
|
|
|
|
// in case of reset.
|
|
|
|
initial_auto_readahead_size_ = initial_auto_readahead_size;
|
2021-12-01 07:52:14 +01:00
|
|
|
readahead_size_ = initial_auto_readahead_size_;
|
2021-04-28 21:52:53 +02:00
|
|
|
readahead_limit_ = 0;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2021-11-11 01:18:27 +01:00
|
|
|
void SetReadaheadState(ReadaheadFileInfo::ReadaheadInfo* readahead_info) {
|
|
|
|
num_file_reads_ = readahead_info->num_file_reads;
|
|
|
|
initial_auto_readahead_size_ = readahead_info->readahead_size;
|
|
|
|
TEST_SYNC_POINT_CALLBACK("BlockPrefetcher::SetReadaheadState",
|
|
|
|
&initial_auto_readahead_size_);
|
|
|
|
}
|
|
|
|
|
De-template block based table iterator (#6531)
Summary:
Right now block based table iterator is used as both of iterating data for block based table, and for the index iterator for partitioend index. This was initially convenient for introducing a new iterator and block type for new index format, while reducing code change. However, these two usage doesn't go with each other very well. For example, Prev() is never called for partitioned index iterator, and some other complexity is maintained in block based iterators, which is not needed for index iterator but maintainers will always need to reason about it. Furthermore, the template usage is not following Google C++ Style which we are following, and makes a large chunk of code tangled together. This commit separate the two iterators. Right now, here is what it is done:
1. Copy the block based iterator code into partitioned index iterator, and de-template them.
2. Remove some code not needed for partitioned index. The upper bound check and tricks are removed. We never tested performance for those tricks when partitioned index is enabled in the first place. It's unlikelyl to generate performance regression, as creating new partitioned index block is much rarer than data blocks.
3. Separate out the prefetch logic to a helper class and both classes call them.
This commit will enable future follow-ups. One direction is that we might separate index iterator interface for data blocks and index blocks, as they are quite different.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6531
Test Plan: build using make and cmake. And build release
Differential Revision: D20473108
fbshipit-source-id: e48011783b339a4257c204cc07507b171b834b0f
2020-03-16 20:17:34 +01:00
|
|
|
private:
|
|
|
|
// Readahead size used in compaction, its value is used only if
|
|
|
|
// lookup_context_.caller = kCompaction.
|
|
|
|
size_t compaction_readahead_size_;
|
|
|
|
|
2021-11-11 01:18:27 +01:00
|
|
|
// readahead_size_ is used if underlying FS supports prefetching.
|
2022-04-16 02:28:09 +02:00
|
|
|
size_t readahead_size_;
|
De-template block based table iterator (#6531)
Summary:
Right now block based table iterator is used as both of iterating data for block based table, and for the index iterator for partitioend index. This was initially convenient for introducing a new iterator and block type for new index format, while reducing code change. However, these two usage doesn't go with each other very well. For example, Prev() is never called for partitioned index iterator, and some other complexity is maintained in block based iterators, which is not needed for index iterator but maintainers will always need to reason about it. Furthermore, the template usage is not following Google C++ Style which we are following, and makes a large chunk of code tangled together. This commit separate the two iterators. Right now, here is what it is done:
1. Copy the block based iterator code into partitioned index iterator, and de-template them.
2. Remove some code not needed for partitioned index. The upper bound check and tricks are removed. We never tested performance for those tricks when partitioned index is enabled in the first place. It's unlikelyl to generate performance regression, as creating new partitioned index block is much rarer than data blocks.
3. Separate out the prefetch logic to a helper class and both classes call them.
This commit will enable future follow-ups. One direction is that we might separate index iterator interface for data blocks and index blocks, as they are quite different.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6531
Test Plan: build using make and cmake. And build release
Differential Revision: D20473108
fbshipit-source-id: e48011783b339a4257c204cc07507b171b834b0f
2020-03-16 20:17:34 +01:00
|
|
|
size_t readahead_limit_ = 0;
|
2021-11-11 01:18:27 +01:00
|
|
|
// initial_auto_readahead_size_ is used if RocksDB uses internal prefetch
|
|
|
|
// buffer.
|
2022-04-16 02:28:09 +02:00
|
|
|
uint64_t initial_auto_readahead_size_;
|
De-template block based table iterator (#6531)
Summary:
Right now block based table iterator is used as both of iterating data for block based table, and for the index iterator for partitioend index. This was initially convenient for introducing a new iterator and block type for new index format, while reducing code change. However, these two usage doesn't go with each other very well. For example, Prev() is never called for partitioned index iterator, and some other complexity is maintained in block based iterators, which is not needed for index iterator but maintainers will always need to reason about it. Furthermore, the template usage is not following Google C++ Style which we are following, and makes a large chunk of code tangled together. This commit separate the two iterators. Right now, here is what it is done:
1. Copy the block based iterator code into partitioned index iterator, and de-template them.
2. Remove some code not needed for partitioned index. The upper bound check and tricks are removed. We never tested performance for those tricks when partitioned index is enabled in the first place. It's unlikelyl to generate performance regression, as creating new partitioned index block is much rarer than data blocks.
3. Separate out the prefetch logic to a helper class and both classes call them.
This commit will enable future follow-ups. One direction is that we might separate index iterator interface for data blocks and index blocks, as they are quite different.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6531
Test Plan: build using make and cmake. And build release
Differential Revision: D20473108
fbshipit-source-id: e48011783b339a4257c204cc07507b171b834b0f
2020-03-16 20:17:34 +01:00
|
|
|
int64_t num_file_reads_ = 0;
|
2021-11-11 01:18:27 +01:00
|
|
|
uint64_t prev_offset_ = 0;
|
2021-04-28 21:52:53 +02:00
|
|
|
size_t prev_len_ = 0;
|
De-template block based table iterator (#6531)
Summary:
Right now block based table iterator is used as both of iterating data for block based table, and for the index iterator for partitioend index. This was initially convenient for introducing a new iterator and block type for new index format, while reducing code change. However, these two usage doesn't go with each other very well. For example, Prev() is never called for partitioned index iterator, and some other complexity is maintained in block based iterators, which is not needed for index iterator but maintainers will always need to reason about it. Furthermore, the template usage is not following Google C++ Style which we are following, and makes a large chunk of code tangled together. This commit separate the two iterators. Right now, here is what it is done:
1. Copy the block based iterator code into partitioned index iterator, and de-template them.
2. Remove some code not needed for partitioned index. The upper bound check and tricks are removed. We never tested performance for those tricks when partitioned index is enabled in the first place. It's unlikelyl to generate performance regression, as creating new partitioned index block is much rarer than data blocks.
3. Separate out the prefetch logic to a helper class and both classes call them.
This commit will enable future follow-ups. One direction is that we might separate index iterator interface for data blocks and index blocks, as they are quite different.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6531
Test Plan: build using make and cmake. And build release
Differential Revision: D20473108
fbshipit-source-id: e48011783b339a4257c204cc07507b171b834b0f
2020-03-16 20:17:34 +01:00
|
|
|
std::unique_ptr<FilePrefetchBuffer> prefetch_buffer_;
|
|
|
|
};
|
|
|
|
} // namespace ROCKSDB_NAMESPACE
|