Amazon Redshift
Cluster Management Guide (API Version 2012-12-01)

Cluster Version History

Find notes about cluster versions for Amazon Redshift.

Versions 1.0.5122, 1.0.5290

Time period: Dec 3–Dec 24, 2018

Features and Improvements

  • Amazon Redshift: Improved performance of hash-based aggregates via more effective memory utilization.

  • Amazon Redshift: Improved performance of queries by elimination of redundant IN subqueries.

  • Amazon Redshift: Improved performance of queries that use outer joins.

  • Amazon Redshift: Improved performance of large cluster resizes.

  • Amazon Redshift: Improved performance for queries that require large hash tables to be created during a hash join operation.

  • Redshift Spectrum: Amazon Redshift now supports dropping an external database by including the DROP EXTERNAL DATABASE clause with a DROP SCHEMA command.

  • Redshift Spectrum: RedshiftSpectrum now supports filtering row-groups in Parquet and ORC files based on the statistics embedded in these files.

Fixes

  • Fix for an issue with UNLOAD to S3 of tables that include varchar columns of maximum length.

  • Fix for an issue where certain queries over a view may fail when the underlying table structure was altered via 'ALTER TABLE ADD COLUMN'.

  • Fix for an issue in PERCENTILE_CONT() window function when computing the 50th percentile.

  • Fix for a query issue where the LIMIT clause was being ignored in certain subqueries.

  • Fix for an issue with handling of NULL values in VARCHAR columns when copying Parquet files.

Versions 1.0.4789, 1.0.4852, 1.0.4896, 1.0.4936

Time period: October 31–Nov 21, 2018

Features and Improvements

  • Amazon Redshift:Amazon Redshift now automatically updates table statistics on your cluster via the new Auto Analyze feature. Up-to-date statistics help the query planner to choose optimal plans and run queries faster. Auto Analyze will be rolling out in all regions in the upcoming releases.

  • Amazon Redshift: Added a DISTSTYLE AUTO distribution style as a default for newly created tables. With DISTSTYLE AUTO, Amazon Redshift now determines optimal distribution style based on table size.

  • Amazon Redshift: VACUUM DELETE now automatically runs in the background to reclaim the space freed by deleted rows. Auto vacuum delete will roll out in all regions in the upcoming releases.

  • Amazon Redshift: Performance improvements for certain hash joins when the inner table join key column is the primary key.

  • Amazon Redshift: Significant performance improvements by using query rewrites that push down selective joins into a subquery

  • Amazon Redshift: Optimize GROUP BY processing by leveraging informational constraints.

  • Amazon Redshift: Added support for cross region UNLOAD with REGION parameter.

  • Amazon Redshift: When the MANIFEST VERBOSE option is used with the UNLOAD command, we now add additional information in the manifest indicating that the author of the manifest is Redshift and the version of the manifest.

  • Amazon Redshift: Added support to cancel an ongoing resize.

  • Amazon Redshift: Update timezone information to Time Zone Database version 2018e.

  • Redshift Spectrum: Query support for nested data has been extended to support arrays of arrays and arrays of maps.

  • Redshift Spectrum: Added support for map data types in Redshift Spectrum to contain arrays.

Fixes

  • Improved memory utilization when many complex views are processed via queries over SVV_COLUMNS.

  • Fix for an issue when using 'orc.schema.resolution' table property with nested external tables.

  • Fix for an issue with INSERT INTO queries that scan nested external tables.

Versions 1.0.4349, 1.0.4515

Time period: October 10–October 31, 2018

Features and Improvements

  • Amazon Redshift: You can now specify the VERBOSE option in the UNLOAD command to generate additional schema and metadata information inside the manifest.

  • Amazon Redshift: You can now manage (both see and cancel) the components of a multi-part query in the AWS console.

  • Amazon Redshift: Significant performance improvements by optimizing the data redistribution strategy during query planning

  • Redshift Spectrum: You can now use Redshift Spectrum with an Amazon Redshift cluster that is configured to use Enhanced VPC Routing.

  • Redshift Spectrum: Improved performance when retrieving metadata for late-binding views.

  • Redshift Spectrum: Improved performance when using ILIKE and NOT ILIKE functions in a Redshift Spectrum query

  • Redshift Spectrum: ALTER TABLE ADD PARTITION now supports creating multiple partitions at once. This makes managing tables with many partitions faster and more convenient.

Fixes

  • Fix for an issue where queries on svv_columns could return an error if there were late-binding views on the cluster.

  • Fix for an issue where certain conversion from TIMESTAMP to CHAR or VARCHAR could truncate milliseconds.

  • Fix for an issue where some very large Redshift Spectrum queries could encounter out of memory errors.

Versions 1.0.3945, 1.0.4081, 1.0.4222

Time period: September 19–October 10, 2018

Features and Improvements

  • Amazon Redshift: You can specify the HEADER option in the UNLOAD command to add a specific header row to each created output file. The header row contains the column names created by the unload query.

  • Amazon Redshift: Significant performance improvements for queries with LIKE predicates.

  • Amazon Redshift: Improved query performance for mixed read and write workloads.

Fixes

  • Enhanced the performance of catalog access for clusters with highly demanding workloads.

  • Fix for an issue when a lateral column alias reference is used with window functions.

  • Fix for an issue with evaluating NULL values in LEAST() and GREATEST() functions.

  • Fix for an issue for certain queries specifying LIMIT 0.

  • Fix for an issue when terminating a session with pg_terminate_backend while the session was returning data.

Versions 1.0.3639, 1.0.3688

Time period: August 29–September 19, 2018

Features and Improvements

  • Amazon Redshift: Significant performance improvements for complex queries that create internal temporary tables during query execution.

  • Amazon Redshift: Significant performance improvements for some queries with rank functions: rank(), dense_rank() and row_number() in a subquery.

  • Redshift Spectrum: Query support for nested data has been extended to support EXISTS and NOT EXISTS subqueries.

  • Redshift Spectrum: Performance improvement of IN-list predicate processing in Redshift Spectrum scans.

Fixes

  • Fix for an issue in queries with a UDF that keeps running beyond the requirements of the LIMIT clause.

  • Fix for a permission issue when a regular view references a late binding view.

Versions 1.0.3324, 1.0.3359

Time period: August 9–August 28, 2018

Features and Improvements

  • Amazon Redshift: Improved performance for write and update workloads by automatically enhancing the locality of system metadata during the maintenance window of your cluster.

  • Amazon Redshift: Significant performance improvements for complex EXCEPT subqueries.

  • Amazon Redshift: Performance improvements for joins involving large numbers of NULL values in a join key column.

  • Amazon Redshift: Performance improvement for queries that refer to stable functions over constant expressions.

  • Amazon Redshift: Performance improvements for queries with intermediate subquery results that can be distributed.

  • Redshift Spectrum: Performance improvements for queries with expressions on the partition columns of external tables.

  • Redshift Spectrum: You can now specify the root of an S3 bucket as the data source for an external table.

Fixes

  • Fix for permission error when the definition of a regular view refers to a late-binding view.

  • Fix for error 'variable not found in subplan target lists' that affected some queries.

Version 1.0.3025

Time period: July 19–August 8, 2018

Features and Improvements

  • Amazon Redshift: Queries can now refer to a column alias within the same query immediately after it is declared, improving the readability of complex SQL queries.

  • Amazon Redshift: DC1 reserved nodes can now be migrated for free to DC2 with no change in term length.

  • Amazon Redshift: A new Amazon CloudWatch metric tracks the current number of waiting queries per workload management (WLM) queue.

  • Amazon Redshift: Made significant performance improvements for single-row inserts into a table.

  • Amazon Redshift: Made significant performance improvements for queries operating over CHAR and VARCHAR columns.

  • Amazon Redshift: Improved performance for hash joins that have filtering dimensions in inner joins or right outer joins.

  • Amazon Redshift: Improved performance for queries that refer to stable functions with constant expressions.

  • Redshift Spectrum: Extended the SVV_COLUMNS catalog view to provide information about the columns of late-binding views.

Fixes

  • Improved memory management for prefetching for wide table scans or aggregations.

  • Improved behavior for long-running queries against Redshift Spectrum external tables with a large number of small files.

Versions 1.0.2762, 1.0.2819, 10.2842

Time period: June 27–July 18, 2018

Features and Improvements

  • Amazon Redshift: Improved handling of short read queries leading to reduced contention and better system utilization.

  • Amazon Redshift: Significant enhancement to the Redshift cluster resize feature, enabling you to create temporary tables during the resize operation.

  • Amazon Redshift: Made further improvements to the compiled code cache, leading to an improvement in overall query performance by reducing the number of segments that have to be recompiled.

  • Amazon Redshift: Performance improvements for the COPY operation when ingesting data from Parquet and ORC file formats.

  • Redshift Spectrum: You can now query external columns during a resize operation.

  • Redshift Spectrum: Made performance improvements for Redshift Spectrum queries with aggregations on partition columns.

Fixes

  • Fix for an issue when reassigning a query from the short query queue to a user queue.

  • Fix for an issue when pushing down predicates during a query rewrite.

  • Fix for an issue with the DATE_TRUNC function in DST timezones.

  • Fix for an issue when the LOWER function is evaluated on specific multibyte character sequences.

  • Fix for a data type derivation issue in a CREATE TABLE AS (CTAS) statement.

  • Enhanced error messaging when the Redshift table total column count limit is exceeded by the number of external tables.

Versions 1.0.2524, 1.0.2557, 1.02610, 1.0.2679, 1.02719

Time period: June 7–July 5, 2018

Features and Improvements

  • Amazon Redshift: Query Monitoring Rules (QMR) now support three times as many rules (up to 25). You can use QMRs to manage the resource allocation of your Amazon Redshift cluster based on query execution boundaries for WLM queues and take action automatically when a query goes beyond those boundaries.

  • Amazon Redshift: A new Amazon CloudWatch metric, QueryRuntimeBreakdown, is now available. You can use this metric to get details on the various query execution stages. For more information, see Amazon Redshift Performance Data.

  • Redshift Spectrum: You can now rename external columns.

  • Redshift Spectrum: You can now specify the file compression type for external tables.

  • Amazon Redshift: Enhanced performance to hash joins when queries involve large joins. Some complex queries now run three times as fast.

  • Amazon Redshift: Substantial enhancements to the VACUUM DELETE command so that it frees up additional disk space and has improved performance.

  • Amazon Redshift: Late materialization now supports DELETE and UPDATE operations, enhancing query performance.

  • Amazon Redshift: Improved performance for cluster resize operations.

  • Redshift Spectrum: Support for an increased number of add and drop operations on a single external Redshift Spectrum table.

  • Redshift Spectrum: Enhanced predicate filtering efficiency when using the DATE_TRUNC function for timestamp columns.

Fixes

  • Fix for an issue for queries based on certain views with constants.

  • Fix for an issue when accessing certain late-binding views after a schema is renamed.

  • Fix for an issue related to unsupported join type.

  • Fix for an issue during query hopping as part of Amazon Redshift Workload Management.

  • Fix for an issue when compiling certain very large queries.

  • Fix for whitespace handling in the format string of the TO_DATE function.

  • Fix for an issue when canceling some queries.

  • Fix an issue when joining on a partition column of CHAR or DECIMAL type.

Version 1.0.2294, 1.0.2369

Time period: May 17–June 14, 2018

Features and Improvements

  • Amazon Redshift: The COPY command now supports loading data in Parquet and ORC columnar formats to Amazon Redshift.

  • Amazon Redshift: Short Query Acceleration now automatically and dynamically determines the maximum run time for short queries based on your workload to further enhance query performance.

  • Redshift Spectrum: ALTER TABLE ADD/DROP COLUMN for external tables is now supported by using standard JDBC calls.

  • Redshift Spectrum: Amazon Redshift can now push the LENGTH() string function to Redshift Spectrum, enhancing performance.

  • Amazon Redshift: Significant improvements to the effectiveness of the compiled code cache leading to an improvement in overall query performance by reducing the number of segments that have to be recompiled.

  • Amazon Redshift: Improved performance of certain UNION ALL queries.

  • Amazon Redshift: Improved performance of the communication layer when dealing with out-of-order network packets.

Fixes

  • Fix for an issue when NULL values are generated for certain queries.

Version 1.0.2058

Time period: April 19–May 8, 2018

Features and Improvements

  • Monitoring functionality and performance enhancements – Amazon CloudWatch metrics for Amazon Redshift provide detailed information about your cluster's health and performance. You can now use the added Query Throughput and Query Latency metrics with Amazon Redshift.

  • Console performance for query plan execution details has been significantly improved – For more information, see Amazon Redshift Performance Data.

  • Amazon Redshift: Enhanced performance for aggregation queries.

  • Amazon Redshift: Added support for closing client connections in case keepalive hasn't been acknowledged.

  • Redshift Spectrum: Support for an increased number of add and drop operations on a single external Redshift Spectrum table.

  • Redshift Spectrum: Enhanced predicate filtering efficiency when using the DATE_TRUNC function for timestamp columns.

Fixes

  • Fix for an issue during the optimization of certain complex queries having an EXISTS clause.

  • Fix for a permission error when some late-binding views are nested within regular views.

  • Fix for an issue during the creation of certain late-binding views that specify 'WITH NO SCHEMA BINDING', where the underlying view definition has a WITH clause.

  • Fix for an issue when deriving data types for columns of certain CREATE TABLE AS SELECT statements involving UNION ALL queries.

  • Fix for Redshift Spectrum issue for DROP TABLE IF EXISTS when the table doesn't exist.

Version 1.0.2044

Time period covered: March 16–April 18, 2018

Features and Improvements

  • Increase to the number of tables you can create to 20,000 for 8xlarge cluster types – You now have greater control and granularity for your data with twice as many tables.

  • New search patterns for REGEXP_INSTR and REGEXP_SUBSTR You can now search for a specific occurrence of a regular expression pattern and specify case sensitivity.

  • Amazon Redshift: Performance of window functions has been significantly improved for 8xlarge clusters.

  • Amazon Redshift: Performance improvement for resize operations.

  • Redshift Spectrum: Now available in Asia Pacific (Mumbai) and South America (São Paulo) regions.

  • Redshift Spectrum: Support for ALTER TABLE ADD/DROP COLUMN for external tables.

  • Redshift Spectrum: New column added to stl_s3query, svl_s3query, and svl_s3query_summary showing file formats for external tables.

Fixes

  • Fix for an issue during query cancellation.

  • Fix for an issue when dropping temp tables.

  • Fix for an issue when altering a table.

Versions 1.0.1792, 1.0.1793, 1.0.1808, 1.0.1865

Time period covered: February 22–March 15, 2018

Features and Improvements

  • Amazon Redshift: Start-time scheduling added to improve the response time of complex queries that are decomposed into multiple subqueries.

  • Redshift Spectrum You can now process scalar JSON and ION file formats for external tables in Amazon S3.

  • Amazon Redshift: Short Query Acceleration: Enhanced accuracy of the ML classifier for highly selective queries.

  • Amazon Redshift: Improved hash join execution time.

  • Amazon Redshift: Improved runtime of vacuum delete operations on DISTALL tables.

  • Redshift Spectrum: Improved runtime for queries of external tables consisting of multiple small files.

  • Redshift Spectrum: Improved performance for querying svv_external_tables and svv_external_columns.

Fixes

  • Improved handling of Boolean predicates during query rewrites.

  • Fix for an issue with inconsistent temp table states.

  • Fix for a race condition between drop and select statements.

  • Enhanced resiliency during cancellation of some queries with prepared statements.

  • Enhanced resiliency for some query rewrites.

Versions 1.0.1691 and 1.0.1703

Time period: February 1–February 23, 2018

Features and Improvements

  • Amazon Redshift: Enhanced support for regular expressions. REGEXP_INSTR() and REGEXP_SUBSTR() can now msearch for a specific occurrence of a regular expression pattern and specify case sensitivity when matching a pattern.

  • Amazon Redshift: Enhanced result caching to support cursor queries and prepared statements.

  • Amazon Redshift: Enhanced Short Query Acceleration to handle a burst of short queries.

  • Amazon Redshift: Added support for IAM role chaining to assume roles that span accounts.

  • Redshift Spectrum: Pushing to Redshift Spectrum the F_TIMESTAMP_PL_INTERVAL, T_CoalesceExpr, and DATE-relates functions.

Fixes

  • Fixes for the calculation of timeouts for very small files on S3.

  • PSQL-c was modified to respect the auto-commit behavior for TRUNCATE

  • Fix for a query compilation failed error.

  • Added support for a three-part column names in late-binding views.

  • Fix for queries with Boolean predicates inside complex expressions.

  • Redshift Spectrum: Increased parallelism using smaller split sizes.

Version 1.0.1636

Time period: January 11–January 25, 2018

Features and Improvements

  • Amazon Redshift: Released Elastic Short Query Acceleration which improves system responsiveness when there is a burst of short queries in a workload.

  • Amazon Redshift: Enabled result caching for read queries within a transaction block.

  • Redshift Spectrum: Added support for DATE columns in Parquet files and for running CREATE EXTERNAL TABLE command with DATE columns.

  • Redshift Spectrum: Added support for IF NOT EXISTS with the ADD PARTITION command.

  • Amazon Redshift: Some short queries over tables with the DISTSTYLE attributes of either KET or ALL now run three times as fast.

Fixes

  • Fix for an intermittent assert error during query processing. The assert message is of the form 'p - id=380 not in rowsets list'

  • Fix for a memory leak during exception handling of certain queries.

  • Addressed inconsistent results between CTAS and INSERT.