Release 0.274

Highlights

Details

General Changes

  • Fix broadcast join memory leak caused by exceeding local memory query failure.

  • Add laplace_cdf() and inverse_laplace_cdf() functions.

  • Reduce the memory footprint and improves the performance of segmented aggregation when the data is already ordered by a subset of the group-by keys. This can be enabled with the segmented_aggregation_enabled session property or the optimizer.segmented-aggregation-enabled configuration property.

  • Add memory limit check for join spilling and fail fast if exceeding to avoid unnecessary processing.

  • Add rate limiting functionality for each query on coordinator http endpoints.

  • Add support for LIKE predicate to SHOW SESSION and SHOW CATALOGS with ESCAPE support.

  • Add support for constraint optimizations to improve performance for querying catalogs that supports table constraints. This feature can be turned on by setting the session property exploit_constraints or config property optimizer.exploit-constraints to true.

  • Add support for local mode on /v1/cluster/ to get information on locally running/queued queries. Can be enabled with the includeLocalInfoOnly property.

  • When cost-based optimizer fails, consider source table size when choosing join order and join distribution type. This might improve query plan for queries with complex projections on small tables. This optimization can be disabled by setting session property size_based_join_distribution_type to false.

SPI Changes

  • Add connector interface serde support to allow customized serde (e.g., thrift) beyond the default JSON ones.

Hive Changes

  • Add support for segmented aggregation to reduce the memory footprint and improve query performance when the order-by keys are a subset of the group-by keys. This can be enabled with the order_based_execution_enabled session property or the hive.order-based-execution-enabled configuration property.

  • Upgrade Hudi support to 0.11.0.

  • Use REQUIRED repetition level for MAP keys in parquet writer.

  • Add decryption functionality to Parquet. When a Parquet file is encrypted following Parquet Modular Encryption, this change enables Presto to be able to decrypt.

Iceberg Changes

  • Add iceberg.max-partitions-per-writer configuration property to allow configuring the limit on partitions per writer.

  • Add support for Nessie catalog.

  • Upgrade Iceberg support to 0.13.2

Mongodb Changes

  • Support renaming tables. Users can use ALTER TABLE RENAME TO syntax to rename a table.

Pinot Changes

  • Fix query error when applying aggregation function over timestamp columns (#17755).

  • Add support for CASE statement pushdown to Pinot.

  • Add support for logical and arithmetic functions for CASE statements.

  • Prefer querying Pinot server grpc endpoint by default (through config pinot.use-streaming-for-segment-queries).

Credits

Ajay George, Amit Adhikari, Amit Dutta, Andy Li, Arjun Gupta, Arunachalam Thirupathi, Beinan, Chunxu Tang, David Simmen, Dongsheng Wang, Eduard Tudenhoefner, Eric Kwok, Guy Moore, Harleen Kaur, Harsha Rastogi, James Sun, Jaromir Vanek, Jonathan Hehir, Ke Wang, Lin Liu, Maria Basmanova, Michael Shang, Nikhil Collooru, Patrick Sullivan, Pranjal Shankhdhar, Rebecca Schlussel, Reetika Agrawal, Rongrong Zhong, Sergey Pershin, Sergii Druzkin, Swapnil Tailor, Tim Meehan, Todd Gao, Xiang Fu, Xiaoman Dong, Xinli shang, Zac, Zhenxiao Luo, abhiseksaikia, ericyuliu, junyi1313, maswin, prithvip, v-jizhang, vaishnavibatni, vigneshwarselvaraj, xiaoxmeng, zhangyanbing