andreacavalli/rocksdb
0
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
rocksdb/db
History
Islam AbdelRahman 68a8e6b8fa Introduce FullMergeV2 (eliminate memcpy from merge operators) 
Summary:
This diff update the code to pin the merge operator operands while the merge operation is done, so that we can eliminate the memcpy cost, to do that we need a new public API for FullMerge that replace the std::deque<std::string> with std::vector<Slice>

This diff is stacked on top of D56493 and D56511

In this diff we
- Update FullMergeV2 arguments to be encapsulated in MergeOperationInput and MergeOperationOutput which will make it easier to add new arguments in the future
- Replace std::deque<std::string> with std::vector<Slice> to pass operands
- Replace MergeContext std::deque with std::vector (based on a simple benchmark I ran https://gist.github.com/IslamAbdelRahman/78fc86c9ab9f52b1df791e58943fb187)
- Allow FullMergeV2 output to be an existing operand

```
[Everything in Memtable | 10K operands | 10 KB each | 1 operand per key]

DEBUG_LEVEL=0 make db_bench -j64 && ./db_bench --benchmarks="mergerandom,readseq,readseq,readseq,readseq,readseq" --merge_operator="max" --merge_keys=10000 --num=10000 --disable_auto_compactions --value_size=10240 --write_buffer_size=1000000000

[FullMergeV2]
readseq      :       0.607 micros/op 1648235 ops/sec; 16121.2 MB/s
readseq      :       0.478 micros/op 2091546 ops/sec; 20457.2 MB/s
readseq      :       0.252 micros/op 3972081 ops/sec; 38850.5 MB/s
readseq      :       0.237 micros/op 4218328 ops/sec; 41259.0 MB/s
readseq      :       0.247 micros/op 4043927 ops/sec; 39553.2 MB/s

[master]
readseq      :       3.935 micros/op 254140 ops/sec; 2485.7 MB/s
readseq      :       3.722 micros/op 268657 ops/sec; 2627.7 MB/s
readseq      :       3.149 micros/op 317605 ops/sec; 3106.5 MB/s
readseq      :       3.125 micros/op 320024 ops/sec; 3130.1 MB/s
readseq      :       4.075 micros/op 245374 ops/sec; 2400.0 MB/s
```

```
[Everything in Memtable | 10K operands | 10 KB each | 10 operand per key]

DEBUG_LEVEL=0 make db_bench -j64 && ./db_bench --benchmarks="mergerandom,readseq,readseq,readseq,readseq,readseq" --merge_operator="max" --merge_keys=1000 --num=10000 --disable_auto_compactions --value_size=10240 --write_buffer_size=1000000000

[FullMergeV2]
readseq      :       3.472 micros/op 288018 ops/sec; 2817.1 MB/s
readseq      :       2.304 micros/op 434027 ops/sec; 4245.2 MB/s
readseq      :       1.163 micros/op 859845 ops/sec; 8410.0 MB/s
readseq      :       1.192 micros/op 838926 ops/sec; 8205.4 MB/s
readseq      :       1.250 micros/op 800000 ops/sec; 7824.7 MB/s

[master]
readseq      :      24.025 micros/op 41623 ops/sec;  407.1 MB/s
readseq      :      18.489 micros/op 54086 ops/sec;  529.0 MB/s
readseq      :      18.693 micros/op 53495 ops/sec;  523.2 MB/s
readseq      :      23.621 micros/op 42335 ops/sec;  414.1 MB/s
readseq      :      18.775 micros/op 53262 ops/sec;  521.0 MB/s

```

```
[Everything in Block cache | 10K operands | 10 KB each | 1 operand per key]

[FullMergeV2]
$ DEBUG_LEVEL=0 make db_bench -j64 && ./db_bench --benchmarks="readseq,readseq,readseq,readseq,readseq" --merge_operator="max" --num=100000 --db="/dev/shm/merge-random-10K-10KB" --cache_size=1000000000 --use_existing_db --disable_auto_compactions
readseq      :      14.741 micros/op 67837 ops/sec;  663.5 MB/s
readseq      :       1.029 micros/op 971446 ops/sec; 9501.6 MB/s
readseq      :       0.974 micros/op 1026229 ops/sec; 10037.4 MB/s
readseq      :       0.965 micros/op 1036080 ops/sec; 10133.8 MB/s
readseq      :       0.943 micros/op 1060657 ops/sec; 10374.2 MB/s

[master]
readseq      :      16.735 micros/op 59755 ops/sec;  584.5 MB/s
readseq      :       3.029 micros/op 330151 ops/sec; 3229.2 MB/s
readseq      :       3.136 micros/op 318883 ops/sec; 3119.0 MB/s
readseq      :       3.065 micros/op 326245 ops/sec; 3191.0 MB/s
readseq      :       3.014 micros/op 331813 ops/sec; 3245.4 MB/s
```

```
[Everything in Block cache | 10K operands | 10 KB each | 10 operand per key]

DEBUG_LEVEL=0 make db_bench -j64 && ./db_bench --benchmarks="readseq,readseq,readseq,readseq,readseq" --merge_operator="max" --num=100000 --db="/dev/shm/merge-random-10-operands-10K-10KB" --cache_size=1000000000 --use_existing_db --disable_auto_compactions

[FullMergeV2]
readseq      :      24.325 micros/op 41109 ops/sec;  402.1 MB/s
readseq      :       1.470 micros/op 680272 ops/sec; 6653.7 MB/s
readseq      :       1.231 micros/op 812347 ops/sec; 7945.5 MB/s
readseq      :       1.091 micros/op 916590 ops/sec; 8965.1 MB/s
readseq      :       1.109 micros/op 901713 ops/sec; 8819.6 MB/s

[master]
readseq      :      27.257 micros/op 36687 ops/sec;  358.8 MB/s
readseq      :       4.443 micros/op 225073 ops/sec; 2201.4 MB/s
readseq      :       5.830 micros/op 171526 ops/sec; 1677.7 MB/s
readseq      :       4.173 micros/op 239635 ops/sec; 2343.8 MB/s
readseq      :       4.150 micros/op 240963 ops/sec; 2356.8 MB/s
```

Test Plan: COMPILE_WITH_ASAN=1 make check -j64

Reviewers: yhchiang, andrewkr, sdong

Reviewed By: sdong

Subscribers: lovro, andrewkr, dhruba

Differential Revision: https://reviews.facebook.net/D57075
2016-07-20 09:49:03 -07:00
..
auto_roll_logger_test.cc
Cleanup auto-roll logger flush-while-rolling test
2016-07-07 11:35:40 -07:00
auto_roll_logger.cc
Eliminate use of 'using namespace std'. Also remove a number of ADL references to std functions.
2016-05-20 07:42:18 -07:00
auto_roll_logger.h
Redo SyncPoints for flush while rolling test
2016-02-22 21:32:19 -08:00
builder.cc
[rocksdb] make more options dynamic
2016-05-17 13:11:56 -07:00
builder.h
[rocksdb] make more options dynamic
2016-05-17 13:11:56 -07:00
c_test.c
Shared dictionary compression using reference block
2016-04-27 17:36:03 -07:00
c.cc
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
column_family_test.cc
Add DestroyColumnFamilyHandle(ColumnFamilyHandle**) to db.h
2016-07-13 17:59:25 -07:00
column_family.cc
Add options.write_buffer_manager: control total memtable size across DB instances
2016-07-05 18:11:25 -07:00
column_family.h
Add options.write_buffer_manager: control total memtable size across DB instances
2016-07-05 18:11:25 -07:00
compact_files_test.cc
Fixed CompactFiles() spuriously failing or corrupting DB
2016-02-22 13:54:58 -08:00
compacted_db_impl.cc
CompactedDB should not be used if there is outstanding WAL files
2016-04-26 14:22:07 -07:00
compacted_db_impl.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
compaction_iterator_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
compaction_iterator.cc
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
compaction_iterator.h
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
compaction_job_stats_test.cc
Introduce XPRESS compresssion on Windows. (#1081)
2016-04-19 22:54:24 -07:00
compaction_job_test.cc
Add options.write_buffer_manager: control total memtable size across DB instances
2016-07-05 18:11:25 -07:00
compaction_job.cc
[rocksdb] make more options dynamic
2016-05-17 13:11:56 -07:00
compaction_job.h
[rocksdb] make more options dynamic
2016-05-17 13:11:56 -07:00
compaction_picker_test.cc
Clean up the ComputeCompactionScore() API
2016-05-23 15:55:29 -07:00
compaction_picker.cc
Update DBTestUniversalCompaction.UniversalCompactionSingleSortedRun to use max_size_amplification_percent = 0
2016-06-27 15:19:27 -07:00
compaction_picker.h
[rocksdb] make more options dynamic
2016-05-17 13:11:56 -07:00
compaction.cc
Update DB::AddFile() to ingest the file to the lowest possible level
2016-06-21 17:57:59 -07:00
compaction.h
Update DB::AddFile() to ingest the file to the lowest possible level
2016-06-21 17:57:59 -07:00
comparator_db_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
convenience.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
corruption_test.cc
Change several option defaults
2016-04-28 17:50:58 -07:00
cuckoo_table_db_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
db_block_cache_test.cc
add option to not flush memtable on open()
2016-06-13 11:34:16 -07:00
db_bloom_filter_test.cc
Deprectate filter_deletes
2016-06-17 10:30:47 -07:00
db_compaction_filter_test.cc
Change some RocksDB default options
2016-03-31 17:12:18 -07:00
db_compaction_test.cc
Add a new feature to enforce a sync point only active on a thread
2016-07-07 11:29:14 -07:00
db_dynamic_level_test.cc
Fix flaky DBTestDynamicLevel.DynamicLevelMaxBytesBase2
2016-05-26 10:13:24 -07:00
db_filesnapshot.cc
Backup Options
2016-06-09 19:03:10 -07:00
db_impl_add_file.cc
update DB::AddFile to ingest list of sst files
2016-07-11 10:43:12 -07:00
db_impl_debug.cc
[rocksdb] make more options dynamic
2016-05-17 13:11:56 -07:00
db_impl_experimental.cc
Clean up the ComputeCompactionScore() API
2016-05-23 15:55:29 -07:00
db_impl_readonly.cc
Add Iterator Property rocksdb.iterator.version_number
2016-03-02 16:23:59 -08:00
db_impl_readonly.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
db_impl.cc
Add DestroyColumnFamilyHandle(ColumnFamilyHandle**) to db.h
2016-07-13 17:59:25 -07:00
db_impl.h
Fix deadlock when trying update options when write stalls
2016-07-12 15:30:38 -07:00
db_info_dumper.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
db_info_dumper.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
db_inplace_update_test.cc
Change some RocksDB default options
2016-03-31 17:12:18 -07:00
db_io_failure_test.cc
Move IO failure test to separate file
2016-05-18 17:09:20 -07:00
db_iter_test.cc
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
db_iter.cc
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
db_iter.h
Introduce PinnedIteratorsManager (Reduce PinData() overhead / Refactor PinData)
2016-04-26 12:41:07 -07:00
db_iterator_test.cc
Fix win build
2016-05-09 11:52:28 -07:00
db_log_iter_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
db_options_test.cc
Fix deadlock when trying update options when write stalls
2016-07-12 15:30:38 -07:00
db_properties_test.cc
Miscellaneous performance improvements
2016-07-12 14:15:32 -07:00
db_sst_test.cc
fix test failure
2016-07-11 13:33:52 -07:00
db_table_properties_test.cc
DBTablePropertiesTest.GetPropertiesOfTablesInRange: Fix Flaky
2016-07-19 15:46:20 -07:00
db_tailing_iter_test.cc
Change Property name from "rocksdb.current_version_number" to "rocksdb.current-super-version-number"
2016-03-04 18:15:29 -08:00
db_test2.cc
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
db_test_util.cc
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
db_test_util.h
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
db_test.cc
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
db_universal_compaction_test.cc
Update DBTestUniversalCompaction.UniversalCompactionSingleSortedRun to use max_size_amplification_percent = 0
2016-06-27 15:19:27 -07:00
db_wal_test.cc
Add options.write_buffer_manager: control total memtable size across DB instances
2016-07-05 18:11:25 -07:00
dbformat_test.cc
Improve BytewiseComparatorImpl::FindShortestSeparator
2016-04-25 23:02:14 -07:00
dbformat.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
dbformat.h
Miscellaneous performance improvements
2016-07-12 14:15:32 -07:00
deletefile_test.cc
Fix a bug that accesses invalid address in iterator cleanup function
2016-07-05 11:57:14 -07:00
event_helpers.cc
Added EventListener::OnTableFileCreationStarted() callback
2016-04-29 11:35:00 -07:00
event_helpers.h
Added EventListener::OnTableFileCreationStarted() callback
2016-04-29 11:35:00 -07:00
experimental.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
fault_injection_test.cc
[rocksdb] Recovery path sequence miscount fix
2016-05-10 14:06:07 -07:00
file_indexer_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
file_indexer.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
file_indexer.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
filename_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
filename.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
filename.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
flush_job_test.cc
Add options.write_buffer_manager: control total memtable size across DB instances
2016-07-05 18:11:25 -07:00
flush_job.cc
Fix clang analyzer errors
2016-07-08 17:50:51 -07:00
flush_job.h
Rename options.compaction_measure_io_stats to options.report_bg_io_stats and include flush too.
2016-04-15 10:22:18 -07:00
flush_scheduler.cc
Fix clang analyzer errors
2016-07-08 17:50:51 -07:00
flush_scheduler.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
forward_iterator_bench.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
forward_iterator.cc
Change Property name from "rocksdb.current_version_number" to "rocksdb.current-super-version-number"
2016-03-04 18:15:29 -08:00
forward_iterator.h
Add Iterator Property rocksdb.iterator.version_number
2016-03-02 16:23:59 -08:00
inlineskiplist_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
inlineskiplist.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
internal_stats.cc
Add InternalStats and logging for AddFile()
2016-06-16 16:21:41 -07:00
internal_stats.h
Add InternalStats and logging for AddFile()
2016-06-16 16:21:41 -07:00
job_context.h
Add More Logging to track total_log_size
2016-07-06 14:29:18 -07:00
listener_test.cc
Add a callback for when memtable is moved to immutable (#1137)
2016-06-02 11:57:31 -07:00
log_format.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
log_reader.cc
Fix clang analyzer errors
2016-07-08 17:50:51 -07:00
log_reader.h
db/log_reader: treat bad record length or checksum as EOF
2016-05-22 22:00:15 -07:00
log_test.cc
db/log_reader: treat bad record length or checksum as EOF
2016-05-22 22:00:15 -07:00
log_writer.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
log_writer.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
managed_iterator.cc
Fixes warnings and ensure correct int behavior on 32-bit platforms.
2016-03-16 22:57:57 +01:00
managed_iterator.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
manual_compaction_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
memtable_allocator.cc
Avoid updating memtable allocated bytes if write_buffer_size is not set
2016-07-13 19:33:57 -07:00
memtable_allocator.h
Add options.write_buffer_manager: control total memtable size across DB instances
2016-07-05 18:11:25 -07:00
memtable_list_test.cc
Add options.write_buffer_manager: control total memtable size across DB instances
2016-07-05 18:11:25 -07:00
memtable_list.cc
group multiple batch of flush into one manifest file (one call to LogAndApply)
2016-07-05 18:09:59 -07:00
memtable_list.h
[rocksdb] Memtable Log Referencing and Prepared Batch Recovery
2016-05-10 14:06:07 -07:00
memtable.cc
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
memtable.h
MemTable::PostProcess() can skip updating num_deletes if the delta is 0
2016-07-19 18:10:18 -07:00
memtablerep_bench.cc
Add options.write_buffer_manager: control total memtable size across DB instances
2016-07-05 18:11:25 -07:00
merge_context.h
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
merge_helper_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
merge_helper.cc
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
merge_helper.h
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
merge_operator.cc
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
merge_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
options_file_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
perf_context_test.cc
CompactedDB should not be used if there is outstanding WAL files
2016-04-26 14:22:07 -07:00
pinned_iterators_manager.h
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
plain_table_db_test.cc
New Statistics to track Compression/Decompression (#1197)
2016-07-19 09:44:03 -07:00
prefix_test.cc
memtable_prefix_bloom_bits -> memtable_prefix_bloom_bits_ratio and deprecate memtable_prefix_bloom_probes
2016-06-10 12:12:10 -07:00
repair_test.cc
ColumnFamilyOptions API [CF + RepairDB part 3/3]
2016-06-24 16:29:43 -07:00
repair.cc
Add options.write_buffer_manager: control total memtable size across DB instances
2016-07-05 18:11:25 -07:00
skiplist_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
skiplist.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
snapshot_impl.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
snapshot_impl.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
table_cache.cc
Enable configurable readahead for iterators
2016-05-04 15:25:58 -07:00
table_cache.h
Enable configurable readahead for iterators
2016-05-04 15:25:58 -07:00
table_properties_collector_test.cc
New Statistics to track Compression/Decompression (#1197)
2016-07-19 09:44:03 -07:00
table_properties_collector.cc
Added "number of merge operands" to statistics in ssts.
2016-05-19 14:24:48 -07:00
table_properties_collector.h
Added "number of merge operands" to statistics in ssts.
2016-05-19 14:24:48 -07:00
transaction_log_impl.cc
Eliminate use of 'using namespace std'. Also remove a number of ADL references to std functions.
2016-05-20 07:42:18 -07:00
transaction_log_impl.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
version_builder_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
version_builder.cc
Adding pin_l0_filter_and_index_blocks_in_cache feature and related fixes.
2016-04-01 10:42:39 -07:00
version_builder.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
version_edit_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
version_edit.cc
Miscellaneous performance improvements
2016-07-12 14:15:32 -07:00
version_edit.h
Avoid FileMetaData copy
2016-07-13 15:36:22 -07:00
version_set_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
version_set.cc
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
version_set.h
Add options.write_buffer_manager: control total memtable size across DB instances
2016-07-05 18:11:25 -07:00
wal_manager_test.cc
Add options.write_buffer_manager: control total memtable size across DB instances
2016-07-05 18:11:25 -07:00
wal_manager.cc
In all the places where log records are read, there was a check that
2016-03-30 23:05:22 +05:30
wal_manager.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
write_batch_base.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
write_batch_internal.h
Deprectate filter_deletes
2016-06-17 10:30:47 -07:00
write_batch_test.cc
Add options.write_buffer_manager: control total memtable size across DB instances
2016-07-05 18:11:25 -07:00
write_batch.cc
Introduce FullMergeV2 (eliminate memcpy from merge operators)
2016-07-20 09:49:03 -07:00
write_callback_test.cc
write_callback_test: clean test directory before running tests
2016-04-18 10:18:41 -07:00
write_callback.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
write_controller_test.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
write_controller.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
write_controller.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
write_thread.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
write_thread.h
fix simple typos (#1183)
2016-06-25 08:29:40 +01:00
xfunc_test_points.cc
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00
xfunc_test_points.h
Updated all copyright headers to the new format.
2016-02-09 15:12:00 -08:00