Make FilterPolicy Customizable #9590
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
pdillinger
approved these changes
Feb 17, 2022
There was a problem hiding this comment.
Generally LGTM. 👍
Modify/expand on this entry in HISTORY.md: "A filter_policy loaded from an OPTIONS file ..." Suggestion:
- Made FilterPolicy Customizable. Configuration of
filter_policyis now accurately saved in OPTIONS file and can be loaded with LoadOptionsFromFile. (Loading an OPTIONS file generated by a previous version only enables reading and using existing filters, not generating new filters. Previously, no filter_policy would be configured from a saved OPTIONS file.)
Of course fix remaining CI issues
Comment on lines
+1316
to
+1326
| const char* BuiltinFilterPolicy::kClassName() { | ||
| return "rocksdb.internal.BuiltinFilter"; | ||
| } | ||
|
|
||
| bool BuiltinFilterPolicy::IsInstanceOf(const std::string& name) const { | ||
| if (name == kClassName()) { | ||
| return true; | ||
| } else { | ||
| return FilterPolicy::IsInstanceOf(name); | ||
| } | ||
| } |
There was a problem hiding this comment.
I'm not sure we need to expose this abstract class in the customizable hierarchy, but OK
Comment on lines
+1362
to
+1372
| const char* BloomLikeFilterPolicy::kClassName() { | ||
| return "rocksdb.internal.BloomLikeFilter"; | ||
| } | ||
|
|
||
| bool BloomLikeFilterPolicy::IsInstanceOf(const std::string& name) const { | ||
| if (name == kClassName()) { | ||
| return true; | ||
| } else { | ||
| return BuiltinFilterPolicy::IsInstanceOf(name); | ||
| } | ||
| } |
There was a problem hiding this comment.
I'm not sure we need to expose this abstract class in the customizable hierarchy, but OK
| @@ -1446,12 +1471,12 @@ FilterBitsBuilder* BloomFilterPolicy::GetBuilderWithContext( | |||
| } | |||
| } | |||
|
|
|||
| const char* BloomFilterPolicy::kName() { return "bloomfilter"; } | |||
| const char* BloomFilterPolicy::kClassName() { return "bloomfilter"; } | |||
There was a problem hiding this comment.
Planning for the future:
For now, the canonical name needs to be "bloomfilter" so that an options file written by 7.0 can be read by many recent versions. But I'm sure we would like to move to another name like "rocksdb.BloomFilter" or something and let "bloomfilter" be the alternate name. So I think to get there, we can add an alternate name now for what we want as the future canonical name, and then in a year or so we can swap the canonical and alternate names.
(Same for ribbonfilter.)
| return guard->get(); | ||
| }); | ||
| library.AddFactory<const FilterPolicy>( | ||
| FilterPatternEntryWithBits(BloomFilterPolicy::kClassName()) |
There was a problem hiding this comment.
Let's add a comment here like
// NOTE: This case previously configured the deprecated block-based filter, but
// old ways of configuring that now map to full filter. We defer to the corresponding
// API to ensure consistency in case that change is reverted.
table/block_based/filter_policy.cc
Outdated
| std::string* /* errmsg */) { | ||
| const std::vector<std::string> vals = StringSplit(uri, ':'); | ||
| double bits_per_key = ParseDouble(vals[1]); | ||
| guard->reset(NewRibbonFilterPolicy(bits_per_key, 0)); |
There was a problem hiding this comment.
I slightly prefer NewRibbonFilterPolicy(bits_per_key) to defer to the corresponding API for second parameter default.
|
@mrambacher has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
pdillinger
reviewed
Feb 18, 2022
table/block_based/filter_policy.cc
Outdated
| .AddSuffix(":false"), | ||
| [](const std::string& uri, std::unique_ptr<const FilterPolicy>* guard, | ||
| std::string* /* errmsg */) { | ||
| // NOTE: This case previously configured the deprecated block-based |
There was a problem hiding this comment.
This belongs with the /*use_block_based_builder=*/true case below
|
@mrambacher has updated the pull request. You must reimport the pull request before landing. |
|
@mrambacher has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
pdillinger
added a commit
to pdillinger/rocksdb
that referenced
this pull request
Mar 22, 2022
Summary: Bloom filters generated by pre-7.0 releases are not read by 7.0.x releases (and vice-versa) due to changes to FilterPolicy::Name() in facebook#9590. This can severely impact read performance and read I/O on upgrade or downgrade with existing DB, but not data correctness. To fix, we go back using the old, unified name in SST metadata but (for a while anyway) recognize the aliases that could be generated by early 7.0.x releases. This unfortunately requires a public API change to avoid interfering with all the good changes from facebook#9590, but the API change only affects users with custom FilterPolicy, which should be very few. Test Plan: manual Generate DBs with ``` ./db_bench.7.0 -db=/dev/shm/rocksdb.7.0 -bloom_bits=10 -cache_index_and_filter_blocks=1 -benchmarks=fillrandom -num=10000000 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 ``` and similar. Compare with ``` for IMPL in 6.29 7.0 fixed; do for DB in 6.29 7.0 fixed; do echo "Testing $IMPL on $DB:"; ./db_bench.$IMPL -db=/dev/shm/rocksdb.$DB -use_existing_db -readonly -bloom_bits=10 -benchmarks=readrandom -num=10000000 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 -duration=10 2>&1 | grep micros/op; done; done ``` Results: ``` Testing 6.29 on 6.29: readrandom : 34.381 micros/op 29085 ops/sec; 3.2 MB/s (291999 of 291999 found) Testing 6.29 on 7.0: readrandom : 190.443 micros/op 5249 ops/sec; 0.6 MB/s (52999 of 52999 found) Testing 6.29 on fixed: readrandom : 40.148 micros/op 24907 ops/sec; 2.8 MB/s (249999 of 249999 found) Testing 7.0 on 6.29: readrandom : 229.430 micros/op 4357 ops/sec; 0.5 MB/s (43999 of 43999 found) Testing 7.0 on 7.0: readrandom : 33.348 micros/op 29986 ops/sec; 3.3 MB/s (299999 of 299999 found) Testing 7.0 on fixed: readrandom : 152.734 micros/op 6546 ops/sec; 0.7 MB/s (65999 of 65999 found) Testing fixed on 6.29: readrandom : 32.024 micros/op 31224 ops/sec; 3.5 MB/s (312999 of 312999 found) Testing fixed on 7.0: readrandom : 33.990 micros/op 29390 ops/sec; 3.3 MB/s (294999 of 294999 found) Testing fixed on fixed: readrandom : 28.714 micros/op 34825 ops/sec; 3.9 MB/s (348999 of 348999 found) ``` Just paying attention to order of magnitude of ops/sec (short test durations, lots of noise), it's clear that with the fix we can read <= 6.29 & >= 7.0 at full speed, where neither 6.29 nor 7.0 can on both. And 6.29 release can properly read fixed DB at full speed.
facebook-github-bot
pushed a commit
that referenced
this pull request
Mar 23, 2022
Summary: Bloom filters generated by pre-7.0 releases are not read by 7.0.x releases (and vice-versa) due to changes to FilterPolicy::Name() in #9590. This can severely impact read performance and read I/O on upgrade or downgrade with existing DB, but not data correctness. To fix, we go back using the old, unified name in SST metadata but (for a while anyway) recognize the aliases that could be generated by early 7.0.x releases. This unfortunately requires a public API change to avoid interfering with all the good changes from #9590, but the API change only affects users with custom FilterPolicy, which should be very few. Pull Request resolved: #9736 Test Plan: manual Generate DBs with ``` ./db_bench.7.0 -db=/dev/shm/rocksdb.7.0 -bloom_bits=10 -cache_index_and_filter_blocks=1 -benchmarks=fillrandom -num=10000000 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 ``` and similar. Compare with ``` for IMPL in 6.29 7.0 fixed; do for DB in 6.29 7.0 fixed; do echo "Testing $IMPL on $DB:"; ./db_bench.$IMPL -db=/dev/shm/rocksdb.$DB -use_existing_db -readonly -bloom_bits=10 -benchmarks=readrandom -num=10000000 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 -duration=10 2>&1 | grep micros/op; done; done ``` Results: ``` Testing 6.29 on 6.29: readrandom : 34.381 micros/op 29085 ops/sec; 3.2 MB/s (291999 of 291999 found) Testing 6.29 on 7.0: readrandom : 190.443 micros/op 5249 ops/sec; 0.6 MB/s (52999 of 52999 found) Testing 6.29 on fixed: readrandom : 40.148 micros/op 24907 ops/sec; 2.8 MB/s (249999 of 249999 found) Testing 7.0 on 6.29: readrandom : 229.430 micros/op 4357 ops/sec; 0.5 MB/s (43999 of 43999 found) Testing 7.0 on 7.0: readrandom : 33.348 micros/op 29986 ops/sec; 3.3 MB/s (299999 of 299999 found) Testing 7.0 on fixed: readrandom : 152.734 micros/op 6546 ops/sec; 0.7 MB/s (65999 of 65999 found) Testing fixed on 6.29: readrandom : 32.024 micros/op 31224 ops/sec; 3.5 MB/s (312999 of 312999 found) Testing fixed on 7.0: readrandom : 33.990 micros/op 29390 ops/sec; 3.3 MB/s (294999 of 294999 found) Testing fixed on fixed: readrandom : 28.714 micros/op 34825 ops/sec; 3.9 MB/s (348999 of 348999 found) ``` Just paying attention to order of magnitude of ops/sec (short test durations, lots of noise), it's clear that with the fix we can read <= 6.29 & >= 7.0 at full speed, where neither 6.29 nor 7.0 can on both. And 6.29 release can properly read fixed DB at full speed. Reviewed By: siying, ajkr Differential Revision: D35057844 Pulled By: pdillinger fbshipit-source-id: a46893a6af4bf084375ebe4728066d00eb08f050
pdillinger
added a commit
to pdillinger/rocksdb
that referenced
this pull request
Mar 23, 2022
…#9736) Summary: Bloom filters generated by pre-7.0 releases are not read by 7.0.x releases (and vice-versa) due to changes to FilterPolicy::Name() in facebook#9590. This can severely impact read performance and read I/O on upgrade or downgrade with existing DB, but not data correctness. To fix, we go back using the old, unified name in SST metadata but (for a while anyway) recognize the aliases that could be generated by early 7.0.x releases. This unfortunately requires a public API change to avoid interfering with all the good changes from facebook#9590, but the API change only affects users with custom FilterPolicy, which should be very few. Pull Request resolved: facebook#9736 Test Plan: manual Generate DBs with ``` ./db_bench.7.0 -db=/dev/shm/rocksdb.7.0 -bloom_bits=10 -cache_index_and_filter_blocks=1 -benchmarks=fillrandom -num=10000000 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 ``` and similar. Compare with ``` for IMPL in 6.29 7.0 fixed; do for DB in 6.29 7.0 fixed; do echo "Testing $IMPL on $DB:"; ./db_bench.$IMPL -db=/dev/shm/rocksdb.$DB -use_existing_db -readonly -bloom_bits=10 -benchmarks=readrandom -num=10000000 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 -duration=10 2>&1 | grep micros/op; done; done ``` Results: ``` Testing 6.29 on 6.29: readrandom : 34.381 micros/op 29085 ops/sec; 3.2 MB/s (291999 of 291999 found) Testing 6.29 on 7.0: readrandom : 190.443 micros/op 5249 ops/sec; 0.6 MB/s (52999 of 52999 found) Testing 6.29 on fixed: readrandom : 40.148 micros/op 24907 ops/sec; 2.8 MB/s (249999 of 249999 found) Testing 7.0 on 6.29: readrandom : 229.430 micros/op 4357 ops/sec; 0.5 MB/s (43999 of 43999 found) Testing 7.0 on 7.0: readrandom : 33.348 micros/op 29986 ops/sec; 3.3 MB/s (299999 of 299999 found) Testing 7.0 on fixed: readrandom : 152.734 micros/op 6546 ops/sec; 0.7 MB/s (65999 of 65999 found) Testing fixed on 6.29: readrandom : 32.024 micros/op 31224 ops/sec; 3.5 MB/s (312999 of 312999 found) Testing fixed on 7.0: readrandom : 33.990 micros/op 29390 ops/sec; 3.3 MB/s (294999 of 294999 found) Testing fixed on fixed: readrandom : 28.714 micros/op 34825 ops/sec; 3.9 MB/s (348999 of 348999 found) ``` Just paying attention to order of magnitude of ops/sec (short test durations, lots of noise), it's clear that with the fix we can read <= 6.29 & >= 7.0 at full speed, where neither 6.29 nor 7.0 can on both. And 6.29 release can properly read fixed DB at full speed. Reviewed By: siying, ajkr Differential Revision: D35057844 Pulled By: pdillinger fbshipit-source-id: a46893a6af4bf084375ebe4728066d00eb08f050
pdillinger
added a commit
to pdillinger/rocksdb
that referenced
this pull request
Mar 23, 2022
…#9736) Summary: Bloom filters generated by pre-7.0 releases are not read by 7.0.x releases (and vice-versa) due to changes to FilterPolicy::Name() in facebook#9590. This can severely impact read performance and read I/O on upgrade or downgrade with existing DB, but not data correctness. To fix, we go back using the old, unified name in SST metadata but (for a while anyway) recognize the aliases that could be generated by early 7.0.x releases. This unfortunately requires a public API change to avoid interfering with all the good changes from facebook#9590, but the API change only affects users with custom FilterPolicy, which should be very few. Pull Request resolved: facebook#9736 Test Plan: manual Generate DBs with ``` ./db_bench.7.0 -db=/dev/shm/rocksdb.7.0 -bloom_bits=10 -cache_index_and_filter_blocks=1 -benchmarks=fillrandom -num=10000000 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 ``` and similar. Compare with ``` for IMPL in 6.29 7.0 fixed; do for DB in 6.29 7.0 fixed; do echo "Testing $IMPL on $DB:"; ./db_bench.$IMPL -db=/dev/shm/rocksdb.$DB -use_existing_db -readonly -bloom_bits=10 -benchmarks=readrandom -num=10000000 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 -duration=10 2>&1 | grep micros/op; done; done ``` Results: ``` Testing 6.29 on 6.29: readrandom : 34.381 micros/op 29085 ops/sec; 3.2 MB/s (291999 of 291999 found) Testing 6.29 on 7.0: readrandom : 190.443 micros/op 5249 ops/sec; 0.6 MB/s (52999 of 52999 found) Testing 6.29 on fixed: readrandom : 40.148 micros/op 24907 ops/sec; 2.8 MB/s (249999 of 249999 found) Testing 7.0 on 6.29: readrandom : 229.430 micros/op 4357 ops/sec; 0.5 MB/s (43999 of 43999 found) Testing 7.0 on 7.0: readrandom : 33.348 micros/op 29986 ops/sec; 3.3 MB/s (299999 of 299999 found) Testing 7.0 on fixed: readrandom : 152.734 micros/op 6546 ops/sec; 0.7 MB/s (65999 of 65999 found) Testing fixed on 6.29: readrandom : 32.024 micros/op 31224 ops/sec; 3.5 MB/s (312999 of 312999 found) Testing fixed on 7.0: readrandom : 33.990 micros/op 29390 ops/sec; 3.3 MB/s (294999 of 294999 found) Testing fixed on fixed: readrandom : 28.714 micros/op 34825 ops/sec; 3.9 MB/s (348999 of 348999 found) ``` Just paying attention to order of magnitude of ops/sec (short test durations, lots of noise), it's clear that with the fix we can read <= 6.29 & >= 7.0 at full speed, where neither 6.29 nor 7.0 can on both. And 6.29 release can properly read fixed DB at full speed. Reviewed By: siying, ajkr Differential Revision: D35057844 Pulled By: pdillinger fbshipit-source-id: a46893a6af4bf084375ebe4728066d00eb08f050
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Make FilterPolicy into a Customizable class. Allow new FilterPolicy to be discovered through the ObjectRegistry