Skip to content

feat: replace semaphore and queued write buffers with pooled approach #446

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
hamersaw wants to merge 2 commits into
base: main
Choose a base branch
from
Open

feat: replace semaphore and queued write buffers with pooled approach #446

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
247c7a8
feat: replace semaphore and queued write buffers with pooled approach
hamersaw Apr 17, 2026
7e73781
fix: use getBufferSizeFor for var-width byte tracking
hamersaw Apr 17, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 5 additions & 10 deletions lance-spark-3.4_2.12/src/main/java/org/lance/spark/write/SparkPositionDeltaWrite.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -212,21 +212,16 @@ private static class PositionDeltaWriteFactory implements DeltaWriterFactory {
@Override
public DeltaWriter<InternalRow> createWriter(int partitionId, long taskId) {
int batchSize = writeOptions.getBatchSize();
boolean useQueuedBuffer = writeOptions.isUseQueuedWriteBuffer();
int poolSize = writeOptions.getQueueDepth();
boolean useLargeVarTypes = writeOptions.isUseLargeVarTypes();
long maxBatchBytes = writeOptions.getMaxBatchBytes();

// Merge initial storage options with write options
WriteParams params = writeOptions.toWriteParams(initialStorageOptions);

// Select buffer type based on configuration
ArrowBatchWriteBuffer writeBuffer;
if (useQueuedBuffer) {
int queueDepth = writeOptions.getQueueDepth();
writeBuffer =
new QueuedArrowBatchWriteBuffer(sparkSchema, batchSize, queueDepth, useLargeVarTypes);
} else {
writeBuffer = new SemaphoreArrowBatchWriteBuffer(sparkSchema, batchSize, useLargeVarTypes);
}
ArrowBatchWriteBuffer writeBuffer =
new PooledArrowBatchWriteBuffer(
sparkSchema, batchSize, poolSize, useLargeVarTypes, maxBatchBytes);

// Create fragment in background thread
Callable<List<FragmentMetadata>> fragmentCreator =
Expand Down
15 changes: 5 additions & 10 deletions lance-spark-3.5_2.12/src/main/java/org/lance/spark/write/SparkPositionDeltaWrite.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -212,21 +212,16 @@ private static class PositionDeltaWriteFactory implements DeltaWriterFactory {
@Override
public DeltaWriter<InternalRow> createWriter(int partitionId, long taskId) {
int batchSize = writeOptions.getBatchSize();
boolean useQueuedBuffer = writeOptions.isUseQueuedWriteBuffer();
int poolSize = writeOptions.getQueueDepth();
boolean useLargeVarTypes = writeOptions.isUseLargeVarTypes();
long maxBatchBytes = writeOptions.getMaxBatchBytes();

// Merge initial storage options with write options
WriteParams params = writeOptions.toWriteParams(initialStorageOptions);

// Select buffer type based on configuration
ArrowBatchWriteBuffer writeBuffer;
if (useQueuedBuffer) {
int queueDepth = writeOptions.getQueueDepth();
writeBuffer =
new QueuedArrowBatchWriteBuffer(sparkSchema, batchSize, queueDepth, useLargeVarTypes);
} else {
writeBuffer = new SemaphoreArrowBatchWriteBuffer(sparkSchema, batchSize, useLargeVarTypes);
}
ArrowBatchWriteBuffer writeBuffer =
new PooledArrowBatchWriteBuffer(
sparkSchema, batchSize, poolSize, useLargeVarTypes, maxBatchBytes);

// Create fragment in background thread
Callable<List<FragmentMetadata>> fragmentCreator =
Expand Down
5 changes: 2 additions & 3 deletions lance-spark-base_2.12/src/main/java/org/lance/spark/write/ArrowBatchWriteBuffer.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,9 +27,8 @@
* Abstract base class for Arrow batch write buffers that bridge Spark row writing and Lance
* fragment creation.
*
* <p>Both {@link SemaphoreArrowBatchWriteBuffer} (lock-based) and {@link
* QueuedArrowBatchWriteBuffer} (queue-based) extend this class, allowing the write path to be
* configured at runtime.
* <p>{@link PooledArrowBatchWriteBuffer} extends this class, pre-allocating a pool of
* VectorSchemaRoots that are reused across batches to minimize allocation overhead.
*/
public abstract class ArrowBatchWriteBuffer extends ArrowReader {

Expand Down
16 changes: 4 additions & 12 deletions lance-spark-base_2.12/src/main/java/org/lance/spark/write/LanceDataWriter.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -131,24 +131,16 @@ protected WriterFactory(
@Override
public DataWriter<InternalRow> createWriter(int partitionId, long taskId) {
int batchSize = writeOptions.getBatchSize();
boolean useQueuedBuffer = writeOptions.isUseQueuedWriteBuffer();
int poolSize = writeOptions.getQueueDepth();
boolean useLargeVarTypes = writeOptions.isUseLargeVarTypes();
long maxBatchBytes = writeOptions.getMaxBatchBytes();

// Merge initial storage options with write options
WriteParams params = writeOptions.toWriteParams(initialStorageOptions);

// Select buffer type based on configuration
ArrowBatchWriteBuffer writeBuffer;
if (useQueuedBuffer) {
int queueDepth = writeOptions.getQueueDepth();
writeBuffer =
new QueuedArrowBatchWriteBuffer(
schema, batchSize, queueDepth, useLargeVarTypes, maxBatchBytes);
} else {
writeBuffer =
new SemaphoreArrowBatchWriteBuffer(schema, batchSize, useLargeVarTypes, maxBatchBytes);
}
ArrowBatchWriteBuffer writeBuffer =
new PooledArrowBatchWriteBuffer(
schema, batchSize, poolSize, useLargeVarTypes, maxBatchBytes);

// Create fragment in background thread
Callable<List<FragmentMetadata>> fragmentCreator =
Expand Down
Loading
Loading