Fix RawBsonDocument encoding performance regression#1888
Merged
vbabanin merged 3 commits intomongodb:mainfrom Feb 24, 2026
Merged
Fix RawBsonDocument encoding performance regression#1888vbabanin merged 3 commits intomongodb:mainfrom
vbabanin merged 3 commits intomongodb:mainfrom
Conversation
Add instanceof check in BsonDocumentCodec and BsonArrayCodec to route RawBsonDocument values to RawBsonDocumentCodec, restoring efficient byte-copy encoding. Previous BsonType-based lookup led to sub-optimal performance as it could not distinguish RawBsonDocument from BsonDocument. JAVA-6101
rozza
added a commit
that referenced
this pull request
Mar 24, 2026
* Revert NettyByteBuf.asReadOnly change (#1871) Originally introduced in 057649f This change had the unintended side effect of leaking netty ByteBufs when logging. JAVA-5982 * Update Netty dependency to the latest version. (#1867) JAVA-5818 * Adjust timeout handling in client-side operations to account for RTT variations (#1793) JAVA-5375 --------- Co-authored-by: Ross Lawley <ross@mongodb.com> * Update Snappy version for the latest security fixes. (#1868) JAVA-6069 * Bson javadoc improvements (#1883) Fixed no comment warning in BinaryVector Improved BsonBinary asUuid documentation Improved BsonBinarySubType isUuid documentation JAVA-6086 * Make NettyByteBuf share parent reference count. (#1891) JAVA-6107 * JAVA-5907 (#1893) * JAVA-5907 * JAVA-5907 use execute within executor service If we don't use the return value from executor then we should use `execute` instead of `submit` * format * revert error log for netty leak --------- Co-authored-by: Almas Abdrazak <abdrazak.almas@mongodb.com> * Fix RawBsonDocument encoding performance regression (#1888) Add instanceof check in BsonDocumentCodec to route RawBsonDocument to RawBsonDocumentCodec, restoring efficient byte-copy encoding. Previous BsonType-based lookup led to sub-optimal performance as it could not distinguish RawBsonDocument from BsonDocument. JAVA-6101 * Update specifications to latest (#1884) JAVA-6092 * Evergreen atlas search fix (#1894) Update evergreen atlas-deployed-task-group configuration Assume test secrets and follow the driver-evergreen-tools atlas recommended usage: https://github.com/mongodb-labs/drivers-evergreen-tools/tree/master/.evergreen/atlas#usage JAVA-6103 * [JAVA-6028] Add Micrometer/OpenTelemetry tracing support to the reactive-streams (#1898) * Add Micrometer/OpenTelemetry tracing support to the reactive-streams driver https://jira.mongodb.org/browse/JAVA-6028 Port the tracing infrastructure from the sync driver to driver-reactive-streams, reusing the existing driver-core, TracingManager, Span, and TraceContext classes. * Move error handling and span lifecycle (span.error(), span.end()) from Reactor's doOnError/doFinally operators into the async callback, before emitting the result to the subscriber. * Making sure span is properly closed when an exception occurs * Clone command event document before storing to prevent use-after-free. (#1901) * Version: bump 5.7.0-beta1 * Version: bump 5.7.0-SNAPSHOT * Remove unneeded variable --------- Co-authored-by: Viacheslav Babanin <slav.babanin@mongodb.com> Co-authored-by: Almas Abdrazak <almas337519@gmail.com> Co-authored-by: Almas Abdrazak <abdrazak.almas@mongodb.com> Co-authored-by: Nabil Hachicha <nabil.hachicha@mongodb.com> Co-authored-by: Nabil Hachicha <1793238+nhachicha@users.noreply.github.com>
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
PR #1632 optimized codec lookup from class-based (
codecRegistry.get(value.getClass())) to BsonType-based (bsonTypeCodecMap.get(value.getBsonType())). This is faster, but it regressed encoding of nestedRawBsonDocumentinsideBsonDocument / BsonArraybecauseRawBsonDocumentandBsonDocumentboth reportBsonType.DOCUMENT. NestedRawBsonDocumentswere encoded viaBsonDocumentCodec, triggering full deserialize and re-serialize instead of thewriter.pipe(reader)byte-copy path.Changes:
instanceof RawBsonDocumentcheck before the BsonType lookup. This keeps the BsonType fast path for the common case while correctly routing the special case forRawBsonDocument.RawBsonNestedEncodingBenchmarkandRawBsonArrayEncodingBenchmarkto CI to prevent regressions for themongot.Why not revert the optimization?
Reverting to class-based lookup would fix correctness but lose the performance gains from PR #1632 (
BsonTypelookup is an array index vsConcurrentHashMaplookup). The instanceof check is negligible and only affects paths withRawBsonDocument.Performance test results
Interpretation: the prior regression is eliminated, and we also see an additional ~13-15% improvement relative to 4.11.x.
JAVA-6101