MaxCompute key feature release notes - MaxCompute - Alibaba Cloud Documentation Center

Major MaxCompute feature releases listed by date, with descriptions and reference links.

All major MaxCompute feature updates are tracked on the Product Updates page.

2025

July

Feature	Description	Release date	Regions	References
MaxCompute supports automatic partitioned tables based on time calculation functions	You can partition a table by applying functions to time or date columns. MaxCompute uses the function results to generate partition values automatically.	2025-07-31	All regions	Automatic partitioned tables based on time calculation functions
MaxCompute supports automatic partitioned tables based on data ingestion time	MaxCompute captures data write timestamps and applies TRUNC_TIME to generate partition column values automatically.	2025-07-31	All regions	Automatic partitioned tables based on data ingestion time
MaxFrame jobs now support worker-level CPU and memory monitoring	MaxFrame lets you use Fuxi Sensor to view the actual CPU and memory consumption of a specific worker.	2025-07-17	All regions	Fuxi Sensor
New built-in large models for MaxFrame AI Function	MaxFrame AI Function now includes built-in Qwen 3 series models, ranging from Qwen3-0.6B to Qwen3-14B.	2025-07-17	All regions	MaxFrame AI Function
MaxFrame now supports partial commit for jobs	In large-scale jobs, some Fuxi instances may run indefinitely due to data skew or infinite loops. Partial commit lets you ignore a small fraction of tasks and retrieve results from successful instances even if the overall job fails.	2025-07-17	All regions	Partial commit for MaxFrame jobs

May

Feature

Description

Release date

Regions

References

MaxCompute SQL engine enhancements

SQL V51 is now available and includes the following updates:

New feature: Support for automatic partitioned tables based on time calculation functions (AUTO PARTITION). You can use time functions to truncate time or date columns and generate values for partition columns.
Syntax enhancements: The STRUCT syntax is enhanced to support the STRUCT(*) syntax. The SHOW HISTORY FOR TASK command now supports specifying a start position and length.
Function enhancements: This version supports new functions, such as BOOL_AND, BOOL_OR, BITWISE_XOR_AGG, and BIT_COUNT.
Performance optimization: Enhanced Shuffle Removal capability improves job performance when writing to clustered tables.

2025-05-30

All regions

Data warehouse engine

April

Feature	Description	Release date	Regions	References
Enhanced observability for MaxCompute data transmission service	This update optimizes metric collection and processing for improved data accuracy. You can now view resource usage at the project level with more options, providing greater observability for the data transmission service.	2025-04-24	China (Hangzhou), China (Shanghai), China (Beijing), China (Shenzhen), China (Chengdu), China East 1 Finance, China (Hong Kong), Singapore, Japan (Tokyo), Malaysia (Kuala Lumpur), Indonesia (Jakarta), Germany (Frankfurt), UK (London), US (Silicon Valley), US (Virginia), and SAU (Riyadh - Partner Region)	Resource observation
MaxFrame AI Function is now available	MaxFrame AI Function provides a programming interface to call large models for offline processing of massive datasets in MaxCompute tables.	2025-04-09	China (Hangzhou), China (Shanghai), China (Beijing), China (Zhangjiakou), China (Ulanqab), China (Shenzhen), China (Chengdu), China (Hong Kong), Japan (Tokyo), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Germany (Frankfurt), UK (London), US (Silicon Valley), US (Virginia), UAE (Dubai), and SAU (Riyadh - Partner Region)	MaxFrame AI Function

March

Feature	Description	Release date	Regions	References
Cost estimation for custom tiered storage policies is now available	You can now estimate storage cost changes for custom tiered storage policies (in addition to the default policy), helping you make informed decisions when configuring policy attributes.	2025-03-24	China (Hangzhou), China (Shanghai), China (Beijing), China (Shenzhen), China (Chengdu), China (Zhangjiakou), China (Ulanqab), China (Hong Kong), Japan (Tokyo), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), and Germany (Frankfurt)	Estimate costs for custom policies
Multi-zone storage is now available	MaxCompute now offers multi-zone storage (the previous default is renamed single-zone storage). Use multi-zone storage for production data to protect against zone-level failures, ensure uninterrupted read/write services, and prevent data loss.	2025-03-24	China (Hangzhou), China (Shanghai), China (Beijing), China (Shenzhen), China East 2 Finance, China (Hong Kong), Singapore, and Indonesia (Jakarta)	Zone-level disaster recovery
Clustering optimization recommendations are now available	Clustering optimization analyzes recent read/write patterns of your tables and suggests clustering configurations to improve job performance and reduce CU consumption. Review estimated benefits, decide whether to apply, and track daily gains afterward.	2025-03-17	China (Hangzhou), China (Shanghai), China (Beijing), China (Zhangjiakou), China (Shenzhen), and China (Chengdu)	Clustering optimization recommendations

February

Feature	Description	Release date	Regions	References
Automatic materialized view (AutoMV) is now available	Automatically creates materialized views based on query patterns to improve computing efficiency and reduce redundant calculations. Ideal when multiple users unknowingly run similar, repetitive queries on MaxCompute tables.	2025-02-20	Public cloud regions in the Chinese mainland	Automatic materialized view (AutoMV)
Public beta for MaxCompute Query Accelerator (MaxQA) 2.0	Built on a dedicated resource pool, MaxQA 2.0 optimizes the control plane, caching, and SQL engine to reduce query response time. Suitable for low-latency scenarios such as BI, interactive analytics, and near real-time data warehousing.	2025-02-17	All regions	MaxCompute Query Accelerator (MaxQA) user guide
MaxCompute SQL engine enhancements	SQL V50 is now available and includes the following updates: Data type and construction enhancements: The DECIMAL data type now supports a scale greater than 18 for higher precision. You can now construct STRUCT data by using STRUCT expressions. Feature enhancements: The FIND_IN_SET function now supports multiple delimiters, and query rewrite for materialized views supports more operators. Performance improvements: This version improves the performance of CTE statements with MATERIALIZED CTE and includes various optimizations for shuffle scenarios.	2025-02-10	All regions	Data warehouse engine
Dynamic data masking is now available	Provides dynamic data masking with multiple policies (masking, hashing, character replacement, rounding for numbers and dates) to protect sensitive data such as PII in development, testing, data sharing, and O&M scenarios.	2025-02-10	China (Hangzhou), China (Shanghai), China (Beijing), China (Zhangjiakou), China (Ulanqab), China (Shenzhen), China (Chengdu), China (Hong Kong), Japan (Tokyo), Malaysia (Kuala Lumpur), Indonesia (Jakarta), and Germany (Frankfurt)	Dynamic data masking
Row-level access control enhancements	This update enhances row-level access control to support more scenarios, such as Result Cache, materialized views, and Local Mode. You can also combine it with other optimization features to improve usability and performance.	2025-02-10	All regions	Row-level access control
Lakehouse and external table enhancements	MaxCompute now implicitly converts data types such as TINYINT, SMALLINT, and DATETIME when you create an external table to parse PARQUET files. You can now use the MAX_PT function to query the most recent partition with data in an OSS external table.	2025-02-10	All regions	Lakehouse and external table capabilities

January

Feature	Description	Release date	Regions	References
The overview page is now available in the console	The overview page is now available in the MaxCompute console. It provides a quick start guide, a list of resource instances, your favorite projects and quotas, alerts and risk warnings, and product updates to streamline your O&M tasks.	2025-01-20	All regions	Project management Quota management Monitoring and alerting

2024

November

Feature	Description	Release date	Regions	References
LogView 2.0 support for MaxFrame	LogView 2.0 shows the execution records, runtimes, and other information for all DAGs submitted in MaxFrame sessions.	2024-11-29	China (Hangzhou), China (Shanghai), China (Beijing), China (Shenzhen), China (Chengdu), China (Ulanqab), and China (Hong Kong)	Use LogView 2.0 to view MaxFrame jobs
Enhancements to materialized view recommendations and management	This update introduces metrics for estimated impact and actual benefits, helping you evaluate recommendations and manage existing materialized views.	2024-11-26	All regions	Recommendations and management of materialized views
Intelligent job diagnostics now available	This feature diagnoses SQL jobs by using detailed execution information to help you quickly identify issues such as job errors or unexpected runtimes and get analysis and troubleshooting suggestions.	2024-11-18	All regions	Intelligent job diagnostics
Automated packaging service for MaxFrame	MaxFrame lets you declare external dependencies during job development, which simplifies managing third-party packages in Python.	2024-11-01	China (Hangzhou), China (Shanghai), China (Beijing), China (Shenzhen), China (Chengdu), China (Ulanqab), and China (Hong Kong)	Automated packaging service

October

Feature	Description	Release date	Regions	References
Support for multiple subscription level-1 quotas in a single region	MaxCompute now allows an Alibaba Cloud account to purchase multiple subscription compute resources in the same region, which creates multiple, isolated level-1 quotas.	2024-10-14	All regions	Subscription Standard

September

Feature

Description

Release date

Regions

References

Storage cost optimization

The storage cost optimization feature recommends tiered storage configuration changes. Based on last access time, it suggests moving data inactive for 30 days to infrequent access (IA) storage and data inactive for 180 days to long-term storage, with estimated cost savings at the list price.

2024-09-27

China (Hangzhou), China (Shanghai), China (Beijing), China (Shenzhen), China (Chengdu), China (Zhangjiakou), and China (Ulanqab)

Optimize tiered storage configuration

Upgraded UI for subscription compute quota management

The UI for managing subscription compute quotas in the MaxCompute console has been upgraded. The management process is now split into two tabs:

Basic Configurations: Use this tab to create and manage level-2 quotas if you do not need time-based quota scaling.
Scaling Configuration: If you need time-based quota scaling, use this tab to configure scalable CU types for your quotas after you set the basic parameters. A new global view shows the CU configurations for all quotas across different time periods.

2024-09-11

All regions

Manage compute quotas

August

Feature	Description	Release date	Regions	References
Storage resource observation now available	You can now view total storage usage and its breakdown by type in a selected region, monitor usage trends, and drill into specific tables or partitions. This helps identify sudden storage increases and optimize high-consumption projects.	2024-08-16	China (Hangzhou), China (Shanghai), China (Beijing), China (Shenzhen), China (Chengdu), China (Zhangjiakou), and China (Ulanqab)	Resource observation
Enhancements to job performance observation	Adds a trend chart for data scanned per CU-hour and top-job lists by resource consumption and runtime. Use these metrics to identify resource-intensive or long-running jobs, pinpoint performance degradation periods, and diagnose root causes.	2024-08-13	All regions	Resource observation

July

Feature	Description	Release date	Regions	References
Job performance observation	Use job performance observation metrics to identify issues like abnormal spikes in job counts or resource usage. Comparing job execution durations over time helps you determine if slow-running jobs are degrading performance.	2024-07-31	All regions	Resource observation
The job O&M list and job analysis overview page now include start time, wait time, and execution duration parameters	These parameters help you quickly determine if long wait times for resources are slowing down your jobs. You can then use job analysis to identify the root cause, such as insufficient resource supply or resource contention among multiple jobs. You can then optimize job execution, adjust job priorities, or manage compute resources to meet your business needs.	2024-07-04	All regions	job O&M
Computing cost optimization is now available in four additional regions	Generates optimal configuration plans for subscription level-1 quotas based on actual job volume and resource requirements. Also previews the impact of switching projects from pay-as-you-go to subscription compute resources to reduce costs and improve utilization.	2024-07-01	China (Hong Kong), Singapore, Indonesia (Jakarta), and Germany (Frankfurt)	Computing resource configuration optimization

June

Feature	Description	Release date	Region	Reference
Row-level access control	Row-level access control lets you define matching rules on a table so users and roles can view only the data they are permitted to access.	2024-06-28	All regions	Row-level access control

May

Feature	Description	Release date	Regions	References
Job performance observation is now available	Use metrics such as job count, CU usage, and job runtime to monitor computing jobs for abnormal spikes in job numbers or resource consumption and to verify that performance meets expectations.	2024-05-08	All regions	Job performance observation

April

Feature	Description	Release date	Region	Reference
Filter clause for window functions	Window functions now support a filter clause to filter out rows unnecessary for window calculations.	2024-04-10	All regions	filter_clause
Support for binary constants	MaxCompute now supports binary constants, which simplifies working with the BINARY data type. You can use the `X'num'` syntax to represent a binary constant.	2024-04-10	All regions	Data types (V2.0)
New built-in function: FROM_CHARSET	MaxCompute `FROM_CHARSET` converts binary data from a specified character set to a UTF-8 string for use in subsequent computations.	2024-04-10	All regions	FROM_CHARSET
New data type: TIMESTAMP_NTZ	The new `TIMESTAMP_NTZ` data type represents a time-zone-unaware timestamp. This simplifies data conversion across different time zones, as its value is independent of the system time zone.	2024-04-10	All regions	TIMESTAMP_NTZ data type

March

Feature	Description	Release date	Regions	References
MaxCompute now available in China (Ulanqab)	MaxCompute, an enterprise-level, serverless, and intelligent data warehouse service, is now available in the China (Ulanqab) region. You can activate the service on demand in the console.	2024-03-26	China (Ulanqab)	-
Store ActionTrail events in MaxCompute	To analyze operational events older than 90 days, you can create a trail in ActionTrail to deliver the events to MaxCompute for querying and analysis.	2024-03-14	All regions	Migrate ActionTrail event data to MaxCompute
SQL Analysis now available in the console	SQL Analysis lets you edit and run SQL commands with basic charting in the MaxCompute console. View metadata for tables, resources, and UDFs across all projects (including tenant-level Information Schema), and test with public datasets.	2024-03-14	All regions	Use SQL Analysis

January

Feature	Description	Release date	Regions	References
Computing cost optimization	The computing cost optimization feature recommends resource configurations and previews the effect of switching pay-as-you-go projects to a subscription, helping you reduce computing costs and improve resource utilization.	2024-01-25	All regions	Optimize computing resource configuration
Customizable metric aggregation in resource observation	Resource observation displays a maximum of 60 data points and defaults to average aggregation when the time range exceeds one hour. You can now set the aggregation to 'maximum' for more comprehensive resource consumption analysis.	2024-01-02	All regions	Resource observation

2023

December

Feature	Description	Release date	Regions	References
Pay-as-you-go Spot Edition: Launched in	MaxCompute now offers the Pay-as-you-go Spot Edition, a cost-effective compute resource for latency-insensitive scenarios.	2023-12-28	China (Hangzhou), China (Shanghai), China (Beijing), China (Shenzhen), China (Hong Kong), Singapore, UK (London), Japan (Tokyo), Malaysia (Kuala Lumpur), Germany (Frankfurt), Indonesia (Jakarta), US (Virginia), and US (Silicon Valley)	Pay-as-you-go Spot Edition
Manage public IP addresses in the console	The MaxCompute console now allows you to add or remove public IP addresses for a project.	2023-12-20	All regions	Configure network access

November

Feature	Description	Release date	Regions	References
Configure dirty data handling for Tablestore external tables	You can now configure dirty data handling when reading from a Tablestore external table.	2023-11-22	All regions	Tablestore external tables
Paimon external table support	MaxCompute now supports Paimon external tables.	2023-11-22	All regions	Paimon external table
Column-level encryption support	Column-level encryption now supports keysets, key rotation, and data encryption and decryption with Key Management Service (KMS) keys.	2023-11-22	All regions	Encryption and decryption functions
Manage the Super Administrator role by using RAM permissions	A RAM user with the `UpdateUsersToSuperAdmin` permission can now assign the Super Administrator role to other project members.	2023-11-21	All regions	Manage projects

September

Feature	Description	Release date	Regions	References
SQL runtime execution mode for MapReduce	MaxCompute now supports running MapReduce jobs in SQL runtime execution mode.	2023-09-25	All regions	SQL runtime execution mode for MapReduce
Job analytics	MaxCompute introduces job analytics to help developers identify and troubleshoot job resource issues.	2023-09-21	All regions	job O&M

August

Feature	Description	Release date	Regions	References
Tenant-level Information Schema	The Tenant-level Information Schema provides metadata, such as project information and usage history. You can retrieve all metadata for a specific object type within a tenant in a single operation.	2023-08-08	All regions	Tenant-level Information Schema

June

Feature	Description	Release date	Regions	References
Support for the JSON data type	MaxCompute supports the JSON data type and automatically optimizes its storage on write, improving computing and analysis performance.	2023-06-27	All regions	JSON data type
Enhanced Z-ordering for data writing	MaxCompute enhances Z-ordering to support global sorting during data writing, reducing the amount of data scanned and improving computing performance.	2023-06-27	All regions	Insert or overwrite data (INSERT INTO \| INSERT OVERWRITE)
Support for clearing column data	You can clear data from specific columns in a table or partition to save storage space.	2023-06-27	All regions	Update or delete data (UPDATE \| DELETE)
Eight new built-in functions for complex types	MaxCompute adds eight new built-in functions for complex types to simplify array operations and improve development efficiency.	2023-06-27	All regions	REVERSE MULTIMAP_FROM_ENTRIES SHUFFLE FLATTEN SEQUENCE ARRAY_NORMALIZE NGRAMS COMBINATIONS

May

Feature	Description	Release date	Regions	References
Write data to Tablestore external tables using the `PutRow` method	MaxCompute now supports writing data to Tablestore external tables using the `PutRow` method.	2023-05-06	All regions	Tablestore external tables
Write data with a header to CSV files in OSS	MaxCompute now lets you specify the first row as the header when you write data to a CSV file in OSS.	2023-05-06	All regions	OSS external tables
New built-in functions	MaxCompute adds four mathematical functions, three date and time functions, and one string function to simplify your SQL code.	2023-05-06	All regions	Mathematical functions ISNAN ATAN2 POSITIVE NEGATIVE Date and time functions DAYOFWEEK DAYOFYEAR CURRENT_TIMEZONE String functions INITCAP
Scheduled data refresh for materialized views	You can schedule data refreshes for materialized views using various methods, ensuring the views contain the latest data.	2023-05-06	All regions	Manage materialized views
Built-in encryption and decryption functions	MaxCompute SQL provides built-in functions to manually encrypt or decrypt data in table columns.	2023-05-06	All regions	Encryption and decryption functions
Filter the results of window functions using `QUALIFY`	MaxCompute now supports the `QUALIFY` clause, which lets you filter the results of window functions based on specified conditions and simplify your SQL code.	2023-05-06	All regions	QUALIFY
Sample table data using `TABLESAMPLE`	MaxCompute now supports the `TABLESAMPLE` clause, which lets you sample table data using different methods.	2023-05-06	All regions	TABLESAMPLE

April

Feature	Description	Release date	Regions	References
Support for schema evolution	You can now modify a table schema by adding columns of complex types, deleting columns, changing column order, and changing column data types.	2023-04-25	All regions	Partition operations
Query acceleration with materialized views	MaxCompute now automatically recommends SQL scripts for creating materialized views. These recommendations help you quickly create materialized views, reduce redundant computations, and save compute resources.	2023-04-23	All regions	Recommendations and management of materialized views
New Job O&M page	On the new Job O&M page, you can view all MaxCompute jobs, including their snapshots and current status.	2023-04-06	All regions	Job O&M

March

Feature	Description	Release date	Regions	References
Process unstructured data in external volumes using Spark and MapReduce tasks	You can use Spark and MapReduce jobs in MaxCompute to process unstructured data in OSS-based external volumes.	2023-03-30	All regions	Process unstructured data using external volumes
Store unstructured data using external volumes	An external volume is a management object in the MaxCompute data warehouse that maps to an OSS path to provide a storage solution for large-scale unstructured data.	2023-03-30	All regions	External volume operations

February

Feature	Description	Release date	Region	References
Enhanced `SHOW` command for data objects	The `SHOW` command now lists data objects by specific categories, including table, view, and function.	2023-02-25	All regions	SHOW

January

Feature	Description	Release date	Region	Related documentation
Direct storage read for Hologres external tables	MaxCompute now supports direct storage reads for Hologres external tables, reducing Hologres connections and improving read performance.	2023-01-21	All regions	Hologres external tables

2022

December

Feature	Description	Release date	Region	Reference
Support for the `WINDOW` keyword in SQL	The `WINDOW` keyword lets you define a window once and reuse it in your queries.	2022-12-14	All regions	WINDOW keyword
Support for the `FROM` clause in `UPDATE` statements	Use the `FROM` clause in an `UPDATE` statement to update data.	2022-12-14	All regions	Update or delete data (UPDATE \| DELETE)
Enhancements to built-in functions	Optimized three built-in string functions: `CAST`, `SPLIT`, and `RAND`. Enhanced two window functions: `NUMERIC_HISTOGRAM` and `PERCENTILE_APPROX`. Added the `MASK_HASH` function.	2022-12-14	All regions	Built-in functions overview

November

Feature	Description	Release date	Regions	References
Support for custom administrative roles	In addition to the built-in `Admin` and `Super_Administrator` roles, MaxCompute now supports custom administrative roles. You can create custom administrative roles with specific permissions, such as managing only permission assignments or IP whitelists, to achieve more granular access control.	2022-11-15	All regions	MaxCompute permissions

October

Feature	Description	Release date	Regions	References
Adds support for upgrading the data structure from `project.table` to `project.schema.table`, simplifying integration with data sources that use a three-level structure.	MaxCompute now supports schemas to better organize objects like tables, resources, and functions within a project. This upgrades the data model from a two-level to a three-level structure. If your source data already uses a three-level hierarchy like `project.schema.table`, you can simplify migration by directly aligning the structures, eliminating the need to refactor your application.	October 13, 2022	All regions	Schema operations

September

Feature	Description	Release date	Region	References
Support for dual-signature authentication for Hologres external tables	The dual-signature mode is an authentication and authorization protocol co-developed by MaxCompute and Hologres. MaxCompute signs authentication data with your logon information and passes it to Hologres for authentication. If you use the same account for both services, you can access Hologres external tables without additional authentication.	2022-09-24	All regions	Hologres external table
Create tables with the same structure as tables in an external data source by using the `CREATE TABLE LIKE` statement.	As a MaxCompute lakehouse feature, you can use the `CREATE TABLE LIKE` syntax to create a MaxCompute table with the same schema as a table in an external data source, such as EMR, Hadoop, or DLF. This feature simplifies schema migration and improves data governance and access performance.	2022-09-23	All regions	Use SQL to manage an external project

August

Feature	Description	Release date	Region	Reference
Automatic deletion of expired partitioned tables	You can configure MaxCompute to automatically delete a partitioned table once data from all of its partitions has been reclaimed according to its lifecycle policy.	2022-08-27	All regions	Lifecycle operations
Three new aggregate functions	Three new aggregate functions are introduced: `BITWISE_AND_AGG` performs a bitwise AND operation on the input values and returns the aggregated value; `MIN_BY` returns the value of a specified column from the row with the minimum value in another specified column; and `MAX_BY` returns the value of a specified column from the row with the maximum value in another specified column.	2022-08-27	All regions	Aggregate functions
Create an internal table by copying the schema of an external table	You can use the `LIKE` clause to copy the schema of an external table when you create an internal table. This makes table creation faster and less error-prone.	2022-08-27	All regions	Create a table
New function to check the status of a materialized view	You can use a new function to check if the data in a materialized view or one of its partitions is consistent with the source table. It returns True for consistent data and False otherwise.	2022-08-27	All regions	Materialized view operations
Support for generating empty partitions in materialized views	When you refresh a partitioned materialized view, MaxCompute generates an empty partition if the refresh operation finds no data for that partition. This ensures partition continuity.	2022-08-27	All regions	Materialized view operations
The "Use Quota" feature for job-level assignment of computing resources	The "Use Quota" feature lets you assign a quota to a specific job, providing more flexible resource management. For example, you can use this feature to isolate computing resources for jobs with different timeliness requirements, such as low-priority backfills or high-priority algorithm jobs. This prevents them from impacting the project's overall performance and avoids the overhead of creating a new project and migrating jobs.	2022-08-23	All regions	Use quotas for computing resources

July

Feature	Description	Release date	Regions	References
New regular expression function	MaxCompute adds the new regular expression function `regexp_extract_all`. This function extracts all matching substrings from a string at once, returning them as an array to improve data processing efficiency.	2022-07-14	All regions	String functions
Customizable file name prefixes and suffixes for UNLOAD	The `UNLOAD` function now lets you set custom prefixes and suffixes for filenames when exporting data from MaxCompute to OSS.	2022-07-14	All regions	UNLOAD
Set the `Split Size` parameter at the table level	You can now set the `Split Size` parameter at the table level to control job parallelism. Adjust this parameter to improve computation efficiency if your jobs run slowly despite sufficient resources or experience long wait times due to resource constraints.	2022-07-14	All regions	SELECT syntax
New window functions and performance optimizations	MaxCompute now supports three new window functions: `first_value`, `last_value`, and `nth_value`. This release also significantly improves the performance of all existing window functions.	2022-07-14	All regions	Window functions
Six new aggregate functions	This release adds six new aggregate functions: `BITWISE_OR_AGG`, `MAP_AGG`, `MULTIMAP_AGG`, `MAP_UNION`, `MAP_UNION_SUM`, and `HISTOGRAM`. You can use these functions to aggregate input bit values or map objects, which simplifies data analysis and statistics.	2022-07-14	All regions	Aggregate functions

June

Feature

Description

Release date

Regions

References

May

Feature	Description	Release date	Regions	References
Independent billing for external table computations	MaxCompute now bills for computations on external tables separately by type. You can see the costs for OSS external tables and Tablestore external tables on your bill, making it easier to track costs from federated computations across different data sources.	2022-05-17	All regions	View bill details

March

Feature	Description	Release date	Regions	References
DISTRIBUTED MAPJOIN support	In specific scenarios, you can use a `DISTRIBUTED MAPJOIN` hint to improve performance and reduce execution time.	2022-03-17	All regions	DISTRIBUTED MAPJOIN
Enhancements to OSS external tables	When you write data to an OSS external table, MaxCompute can automatically create a managed directory. You can also specify the cache capacity for reading files when you create an OSS external table.	2022-03-17	All regions	OSS external table
New parsing method for JSON data	You can now parse JSON keys that contain a period (`.`) by enclosing them in brackets and single quotation marks (`['']`).	2022-03-17	All regions	GET_JSON_OBJECT_TUPLE and JSON_TUPLE
Enhancements to TRIM, LTRIM, and RTRIM functions	The `TRIM`, `LTRIM`, and `RTRIM` functions can now remove specified characters from the left, right, or both ends of a string.	2022-03-17	All regions	string functions
Enhancements to automatic query rewrite for materialized views	Automatic query rewrite for materialized views now supports `OUTER JOIN`, `UNION`, and `UNION ALL`.	2022-03-17	All regions	materialized view operations
Skip headers and footers in compressed TEXTFILE files	You can now use the `skip.header.line.count` and `skip.footer.line.count` properties to skip header and footer rows in TEXTFILE files. This feature is supported for both uncompressed files and files compressed in GZIP, BZIP2, or LZO format.	2022-03-01	All regions	ORC external table
Compatibility with Apache Spark 3.1	MaxCompute now supports Apache Spark 3.1, in addition to versions 1.6, 2.3, and 2.4.	2022-03-01	All regions	Set up a Linux development environment

February

Feature	Description	Release date	Regions	References
Data security enhancements in MaxCompute Logview	You can now control the visibility of job execution results in Logview to enhance data security.	2022-02-25	All regions	Project operations
Schema evolution in MaxCompute	Schema evolution now supports adding columns of complex data types, deleting columns, and reordering columns.	2022-02-23	All regions	Partition operations

January

Feature	Description	Release date	Region	Reference
View MaxCompute external project metadata in DataWorks Data Map	You can view metadata from MaxCompute external projects in DataWorks Data Map.	2022-01-10	Germany (Frankfurt) and Singapore	Manage a lakehouse in DataWorks

2021

December

Feature	Description	Release date	Regions	References
Wizard-driven setup for MaxCompute connections to Hadoop Hive in a VPC, Data Lake Formation (DLF), and Object Storage Service (OSS)	MaxCompute now provides a wizard-driven process to create a network connection to a VPC, a data source connection, and an external project for your lakehouse. This process replaces the previous ticket-based method, reducing setup time from days to minutes. Additionally, a new lakehouse page centralizes lakehouse creation, development, management, and governance.	2021-12-30	All regions	Overview of the MaxCompute lakehouse
Progressive computing in MaxCompute	MaxCompute now supports progressive computing. This feature automatically saves intermediate results into time-based partitions at a specified time granularity. Subsequent job runs can reuse previously computed results from overlapping time periods. This conserves compute resources, shortens job scheduling time, lowers costs, and improves overall job efficiency.	2021-12-03	All regions	Progressive computing
Materialized view enhancements in MaxCompute	You can now create partitions and clusters on materialized views. If a query requests data from a partition unavailable in the materialized view, you can configure the system to automatically fall back to the source table. The system then returns a combined result from the materialized view and the source table.	2021-12-01	All regions	Materialized view operations

November

Feature	Description	Release date	Regions	References
Manage VPC connections in the console	MaxCompute can access services in a VPC using features such as external tables, user-defined functions (UDFs), and lakehouse. To access these services, you must establish a network connection between MaxCompute and the target service, such as an ApsaraDB for HBase, ApsaraDB RDS, or a Hadoop cluster. Previously, setting up a VPC connection required submitting a form and waiting for approval. Now, you can add or delete network connections on the Network Resources page in the MaxCompute console. This self-service process improves development efficiency. Before you begin, ensure your account has the required permissions to manage Network Resources.	2021-11-29	China (Beijing) China (Shanghai) China (Zhangjiakou) China (Hangzhou) China (Shenzhen)	Configure network access
Clear data from specified partitions in a partitioned table	You can now manually clear data from specific partitions in a partitioned table without deleting the table or its partition metadata. You can also use filter conditions to select the partitions to clear.	2021-11-23	All regions	Clear partition data

October

Feature	Description	Release date	Region	References
Support for ApsaraDB for HBase Performance-enhanced Edition and Lindorm via MaxCompute external tables	You can now create external tables in MaxCompute to read from and write to ApsaraDB for HBase Performance-enhanced Edition and Lindorm. This feature lets you perform data processing, analysis, and federated queries across these services. ApsaraDB for HBase Performance-enhanced Edition is a cloud-hosted database that is fully compatible with HBase and powered by the wide table engine of Lindorm. Lindorm is a cloud-native, multi-model database designed for scenarios such as IoT, the internet, and the Internet of Vehicles (IoV). It unifies access to and processing of various data types, including wide tables, time series, text, objects, streams, and spatial data.	2021-10-29	All regions	Lindorm external table

September

Feature	Description	Release date	Region	Related documents
MaxCompute launches in the China (Shenzhen) Finance region	MaxCompute is now available in the China (Shenzhen) Finance region.	2021-09-14	China (Shenzhen) Finance	None
Access Kerberos-enabled Hadoop clusters	Most enterprise production Hadoop clusters use Kerberos for authentication and authorization. MaxCompute can now access these clusters, which significantly expands its lakehouse integration with Hadoop.	2021-09-01	China (Hangzhou) China (Shanghai) China (Beijing) China (Shenzhen) China (Zhangjiakou) Singapore	None
Read Hudi and Delta Lake data from OSS with the lakehouse solution	Delta Lake and Hudi are storage mechanisms commonly used in data lake solutions for both stream and batch processing. MaxCompute provides a lakehouse architecture built on Alibaba Cloud DLF and OSS that supports these storage mechanisms. You can use MaxCompute to query real-time data for immediate insights into business data changes.	2021-09-01	China (Hangzhou) China (Shanghai) China (Beijing) China (Shenzhen) Singapore	Support for the Delta Lake and Hudi storage mechanisms based on DLF, RDS, or Flink, and OSS

August

Feature	Description	Release date	Region	References
MaxCompute SQL adds 16 new built-in functions	The new built-in functions in MaxCompute SQL include: Complex type functions FIELD: Retrieves the value of a member variable from a STRUCT. Date and time functions TO_MILLIS: Converts a specified date to a UNIX timestamp in milliseconds. String functions ENCODE: Encodes a string in a specified encoding format. FIND_IN_SET: Finds the position of a specified string within a comma-separated string. LOCATE: Returns the position of a substring within a string. PARSE_URL_TUPLE: Parses a URL and returns multiple URL components. Mathematical functions CORR: Calculates the Pearson correlation coefficient for two columns of data. Other functions HASH: Computes a hash value for the input parameters. COMPRESS: Compresses an input string using the GZIP algorithm. DECOMPRESS: Decompresses BINARY input using the GZIP algorithm. NULLIF: Returns NULL if expr1 is equal to expr2. Otherwise, returns expr1. FORMAT_NUMBER: Converts a number to a string in a specified format. SHA: Calculates the SHA-1 hash of a string or binary value and returns it as a hexadecimal string. SHA1: Calculates the SHA-1 hash of a string or binary value and returns it as a hexadecimal string. SHA2: Calculates the SHA-2 family hash of a string or binary value. Supported algorithms include SHA-224, SHA-256, SHA-384, and SHA-512. CRC32: Calculates the cyclic redundancy check (CRC) of a string or binary value.	2021-08-31	All regions	Complex type functions Date and time functions String functions Mathematical functions Other functions
UPDATE, DELETE, and MERGE INTO in MaxCompute SQL are now generally available	The public preview for the UPDATE, DELETE, and MERGE INTO statements in MaxCompute SQL ended on August 10, 2021. These features are now generally available and subject to billing. Pay-as-you-go jobs that use these statements will now incur computing charges. If you previously hesitated to use these features due to stability concerns during the public preview, you can now use them with confidence. These features now have the same availability and stability guarantees as other MaxCompute SQL capabilities.	2021-08-10	China (Beijing) China (Shanghai) China (Zhangjiakou) China (Hangzhou) China (Shenzhen) China (Chengdu)	Update or delete data (UPDATE \| DELETE) MERGE INTO
MaxCompute Streaming Tunnel is now generally available	MaxCompute Streaming Tunnel provides streaming semantic APIs to simplify the development of distributed services. It supports concurrent partition creation and asynchronous data sorting using `zorder by`.	2021-08-09	All regions	Streaming Tunnel
MaxCompute SQL supports materialized views	MaxCompute SQL introduces materialized views. A materialized view is a database object that stores the precomputed results of a query. MaxCompute accelerates subsequent queries by reusing results from time-consuming operations, such as joins and aggregations. Materialized views are ideal for the following scenarios: Frequently executed queries with fixed patterns. Queries that involve time-consuming operations, such as aggregations and joins. After you create a materialized view, MaxCompute automatically rewrites incoming queries to use the optimal materialized view. It then reads data directly from the view, significantly improving query efficiency without requiring you to modify your existing queries.	2021-08-06	All regions	Materialized view operations

July

Feature	Description	Release date	Regions	References
Support for pre-sorted input to aggregate functions in MaxCompute SQL	For the`WM_CONCAT` and`COLLECT_LIST` aggregate functions, and user-defined aggregate functions (UDAFs) that are sensitive to the input order, use the`WITHIN GROUP (ORDER BY col1[, col2...])` clause to sort the input.	2021-07-30	All regions	Aggregate functions
Support for multi-column operations in SQL subqueries	MaxCompute SQL is now compatible with PostgreSQL subquery syntax, enabling multi-column operations in`IN` subqueries and scalar subqueries. For`IN` subqueries: The expression after`IN` can be a simple multi-column SELECT statement. In the expression that follows `in`, you can use aggregate functions. The expression after`IN` can be a constant. `scalar subquery`: The SELECT list can contain a multi-column scalar subquery, but only in an equality expression. The SELECT list can contain a BOOLEAN expression, but only for an equality comparison. The`WHERE` clause supports multi-column comparison, but only for equality.	2021-07-29	All regions	Subqueries
Enhanced project deletion options	You can delete a MaxCompute project in the MaxCompute console using an Alibaba Cloud account or a RAM user with the Super_Administrator role for the project. Two deletion methods are available: Logically Delete and Allow Project Restoration Within 15 Days: The project becomes unavailable after deletion but can be restored within 15 days. To restore the project, go to the Project management tab, find the target MaxCompute project, and click Restore in the Actions column. If you do not restore it, MaxCompute permanently deletes the project after 15 days. Immediately Delete and Prohibit Project Restoration: The project is permanently deleted and cannot be restored. You can immediately create a new project with the same name.	2021-07-29	China (Beijing) China (Shanghai) China (Zhangjiakou) China (Hangzhou) China (Shenzhen) China (Chengdu)	Delete a table or MaxCompute project

April

Feature	Description	Release date	Region	References
Dynamic filtering optimization	In distributed systems, joining large datasets is time-consuming and resource-intensive. To improve performance, MaxCompute introduces dynamic filtering, an optimization that generates runtime filters from a join's equi-join conditions. These filters accelerate queries by pruning data before the shuffle and join phases. This optimization is particularly effective for queries that join a dimension table with a fact table. You can enable dynamic filtering at the session level with a switch, trigger it manually with hints, or allow MaxCompute to apply it automatically. To further improve performance, use this feature with dynamic partition pruning.	2021-04-07	China (Beijing) China (Shanghai) China (Zhangjiakou) China (Hangzhou) China (Shenzhen) China (Chengdu)	Dynamic Filtering

March

Feature	Description	Release date	Region	Documentation
Timeout monitoring and alerts for all jobs or all SQL jobs in a MaxCompute project	MaxCompute lets you use CloudMonitor to configure threshold-based alert rules to monitor job runtimes. When a job times out, the system sends an alert to the specified contacts. This helps you promptly identify abnormal jobs and improve O&M efficiency.	2021-03-16	China (Beijing) China (Shanghai) China (Zhangjiakou) China (Hangzhou) China (Shenzhen) China (Chengdu)	Job timeout monitoring and alerts
MaxCompute SQL supports `FROM_JSON` and `TO_JSON` functions	The `FROM_JSON` function can convert a JSON string into a MaxCompute data type or extract information from the JSON string. Based on the JSON string `jsonStr` and the schema information, this function returns a value of the `ARRAY`, `MAP`, or `STRUCT` type. The `TO_JSON` function converts a complex type, such as `ARRAY`, `MAP`, or `STRUCT`, into a JSON string.	2021-03-16	all regions	String functions
MaxCompute SQL supports `UPDATE`, `DELETE`, and `MERGE INTO` syntax (public preview)	`UPDATE` and `DELETE` are new MaxCompute statements for row-level operations on table or partition data. Previously, to correct even a small amount of data in a table or partition, you had to read all the data, generate new data, and then use `INSERT OVERWRITE` to write it all back. The `UPDATE` and `DELETE` operations significantly reduce the amount of data read and written for these tasks. When your operation on a target table includes `INSERT`, `UPDATE`, and `DELETE` logic, the `MERGE INTO` feature lets you combine these operations into a single, more efficient SQL statement. Additionally, `MERGE INTO` jobs are atomic. If any internal logic fails, the entire job fails. This prevents partial updates because the entire job succeeds or fails as a single unit. You must create a transactional table to use the `UPDATE`, `DELETE`, and `MERGE INTO` features. Because big data systems guarantee job-level ACID semantics, job conflicts may occur when jobs run concurrently and operate on the same target table. Note that `UPDATE`, `DELETE`, and `MERGE INTO` are currently in public preview. No computing fees are charged during the public preview. However, during this period, we do not provide guarantees for production jobs or data that use this feature. Back up your important data.	2021-03-16	all regions	Update or delete data (`UPDATE` \| `DELETE`) MERGE INTO Table operations ACID semantics

February

Feature	Description	Release date	Region	References
The MaxCompute lakehouse solution is now generally available	The MaxCompute lakehouse solution integrates a MaxCompute data warehouse with a data lake, such as OSS or Hadoop HDFS. You can build a lakehouse in two ways: Build a lakehouse with MaxCompute, OSS, and Data Lake Formation (DLF): This method uses DLF to manage the metadata of your data lake. MaxCompute leverages the metadata management capabilities of DLF to efficiently process semi-structured OSS data in formats such as AVRO, CSV, JSON, PARQUET, and ORC. Build a lakehouse with MaxCompute and HDFS: This method integrates MaxCompute with your HDFS cluster, whether it is self-managed on-premises, deployed on cloud virtual machines (VMs), or part of an Alibaba Cloud E-MapReduce cluster. To use this solution, submit an application. The MaxCompute team will contact you to help with the setup.	2021-02-26	China (Beijing) China (Shanghai) China (Hangzhou)	Overview of the MaxCompute lakehouse
Support for ApsaraDB for HBase external tables (public preview)	MaxCompute can now connect to ApsaraDB for HBase over a VPC. After you establish the network connection and grant permissions, you can create ApsaraDB for HBase external tables to read from and write to tables in an ApsaraDB for HBase database by using the HBaseStorageHandler interface provided by Hive. This allows you to synchronize data from ApsaraDB for HBase to MaxCompute for extract, transform, and load (ETL) operations, perform federated computing, or export data from MaxCompute back to ApsaraDB for HBase. This public preview feature is available free of charge and without a Service Level Agreement (SLA).	2021-02-08	China (Beijing) China (Shanghai) China (Zhangjiakou) China (Hangzhou)	ApsaraDB for HBase external tables (Standard Edition or Performance-enhanced Edition)
Disable header display in the MaxCompute command-line tool	When using a shell script to retrieve dynamic values from `odpscmd -e SQL_statement`, you may want the output to contain only data, without headers or other runtime information. You can now use the `set odps.sql.select.output.format={needHeader:false,fieldDelim:""};` command to disable headers and get clean, data-only output.	2021-02-08	All regions	Connect by using the local client (odpscmd)
Enhanced data writing to OSS external tables	MaxCompute now uses the OSS multipart upload feature to more efficiently write data to OSS external tables during INSERT operations. Additionally, data that MaxCompute writes to an OSS external table is stored in a `.odps` folder within the LOCATION directory, and maintains a `.meta` file to ensure data consistency. Only MaxCompute can correctly process the content in the `.odps` folder. Errors may occur if other engines read this content. MaxCompute introduces the `odps.sql.unstructured.oss.commit.mode` property. When this property is set to `true` (the default value is `false`), MaxCompute uses the multipart upload feature and a `two-phase commit` method to ensure data consistency. In this mode, the `.odps` directory and the `.meta` file are not created. This ensures compatibility and allows other data processing engines to read the data.	2021-02-08	All regions	Write data to OSS
Support for custom SerDes in Hive-compatible external tables	MaxCompute is compatible with the Hive SerDe interface and provides built-in SerDe classes for common open source data formats. For special data formats, you can now specify a custom `ROW FORMAT SERDE`. To use a custom SerDe, add its JAR package as a resource and include the `using jar` clause in the `CREATE EXTERNAL TABLE` statement.	2021-02-08	All regions	Support for open source data formats
New and enhanced built-in functions	The following functions are added or enhanced: The `date`, `datediff`, `datepart`, and `datetrunc` date functions now support the DATE and TIMESTAMP data types. New string functions include `parse_url` for parsing URLs, and `base64` and `unbase64` for converting data between binary and Base64-encoded strings. Aggregate functions now support a filter clause, allowing you to specify a filter condition for a specific aggregate function in a `select` clause. A new `count_if` function counts records that meet a specified condition. Other new functions include `stack` to split data into multiple rows for compatibility with Hive and Spark SQL; `get_user_id` to retrieve the current user's ID; and `array_intersect` to find the common elements in two `array`s. The `sort_array` function is also enhanced with an isAsc parameter to specify the sort order (ascending by default).	2021-02-08	All regions	Date and time functions String functions Aggregate functions Other functions
New optimizer metadata collection tools: Freeride and Analyze	MaxCompute now provides two ways to collect table statistics for the optimizer: Asynchronous collection (Analyze): You can run commands to proactively collect metadata from a specified table. This operation incurs fees for scanning the table data. Synchronous collection (Freeride): You can enable Freeride by setting parameters before a `create table ... as ...` or `insert` statement. This method automatically collects column statistics as data is generated. This method is more automated but may increase query latency.	2021-02-08	All regions	Optimizer
Support for ApsaraDB RDS external tables (public preview)	MaxCompute can now connect to your ApsaraDB RDS instance over a VPC. After you establish the network connection and grant permissions, you can create ApsaraDB RDS external tables to read and write data. This allows you to synchronize data, such as dimension or business data, from ApsaraDB RDS to MaxCompute for ETL, perform federated computing, or export data from MaxCompute back to your ApsaraDB RDS database. This public preview feature is available free of charge and without an SLA.	2021-02-08	All regions	ApsaraDB RDS for MySQL external tables
Read from and write to Hologres external tables by using JDBC (public preview)	MaxCompute external tables can now access Hologres data sources using a JDBC driver. You can create a Hologres external table in MaxCompute and use the PostgreSQL JDBC driver to query data in your Hologres instance. This method eliminates redundant storage and data import/export steps, accelerating query performance. This public preview feature is available free of charge and without an SLA.	2021-02-08	All regions	Hologres external tables
Export data to OSS by using the UNLOAD command	MaxCompute now supports the UNLOAD command to export data to OSS. This allows you to store structured data in OSS and use other computing engines to process and analyze the data.	2021-02-08	All regions	UNLOAD
MaxCompute SQLML is now generally available	MaxCompute SQLML allows data engineers, analysts, and data scientists to use SQL to create, train, and apply machine learning models directly in MaxCompute. This enables SQL practitioners to leverage machine learning capabilities with their existing tools and skills, without migrating data.	2021-02-01	All regions	Overview Quick Start Billing and metering

January

Feature	Description	Release date	Region	References
MaxCompute storage pricing adjustment	Effective January 10, 2021, the MaxCompute storage service switched from tiered pricing to a fixed rate. This change reduces storage costs for customers with small to medium-sized datasets. The new fixed rate is USD 0.018 per GB per month (equivalent to USD 0.0006 per GB per day).	2021-01-10	all regions	Storage pricing

2020

December

Feature	Description	Release date	Region	References
Information Schema is now enabled by default for new projects	Information Schema is now enabled by default in all new projects, eliminating the need for manual setup and allowing you to easily view metadata.	2020-12-21	All regions	Project-level Information Schema
Information_Schema.Tasks view for running job snapshots is now in private preview	The Information_Schema.Tasks view details currently running jobs, including the job list, CPU and memory usage, resource consumption ratio, and runtime. This view helps you quickly identify long-running or resource-intensive jobs for targeted optimization or resource capacity planning.	2020-12-08	None.	Metadata views
Query acceleration is now supported in the MaxCompute query editor	The MaxCompute query editor now supports query acceleration, which reduces the execution time for queries on small to medium-sized datasets from minutes to seconds. When you submit a query, the editor first routes it to the query acceleration service. If the query meets the acceleration criteria, the service returns results in seconds. If not, the query automatically falls back to the standard offline execution mode. The query editor provides rich web-based analysis capabilities for querying data, analyzing results, and sharing insights in one place. Query acceleration is primarily available for pay-as-you-go resources and is not yet supported for projects that use subscription resources.	2020-12-05	All regions	Connect by using DataWorks

November

Feature	Description	Release date	Region	References
Continuously import data from Message Queue for Apache Kafka using a sink connector	MaxCompute is tightly integrated with Message Queue for Apache Kafka. The sink connector lets you continuously import data from a specified topic into a MaxCompute table without needing third-party tools or custom code. This feature simplifies the data integration pipeline and reduces development and O&M costs.	2020-11-27	All regions	Import data to MaxCompute with Kafka (offline and real-time)
Access MaxCompute using a RAM role	In addition to Alibaba Cloud accounts and RAM users, MaxCompute projects now support access through RAM roles. After an administrator adds a RAM role to a project and grants permissions, users can assume the role to access MaxCompute. This method uses temporary STS credentials for secure access without a long-term AccessKey. Because the MaxCompute console supports STS authentication, you can use a RAM role for data analysis and development.	2020-11-24	All regions	User authentication

June

Feature	Description	Release date	Region	References
Data type version selection for new projects	When you create a project in MaxCompute, you must specify a data type version.	2020-06-30	All regions	Data type versions

April

Feature	Description	Release date	Region	References
policy authorization with the `GRANT` statement	MaxCompute supports policy authorization with the `GRANT` command to simplify the authorization process.	2020-04-23	all regions	policy-based access control

March

Feature	Description	Release date	Region	References
Five new built-in functions	MaxCompute now supports five new built-in functions: `TABLE_EXISTS()`, `PARTITION_EXISTS()`, `NUMERIC_HISTOGRAM()`, `PERCENTILE_APPROX()`, and `FORMAT_NUMBER()`.	2020-03-17	all regions	other functions aggregate functions mathematical functions

February

Feature	Description	Release date	Region	References
Disable table lifecycles	You can use the `DISABLE LIFECYCLE` parameter to disable the lifecycle of a table.	2020-02-26	All regions	Lifecycle operations
New `CLONE TABLE` command	The `CLONE TABLE` command copies data from one table to another.	2020-02-26	All regions	CLONE TABLE
Upgrading `GROUPING SETS`	MaxCompute extends `GROUPING SETS` to support nesting `CUBE`, `ROLLUP`, and `GROUPING SETS`.	2020-02-26	All regions	GROUPING SETS
Fetch external table data over HTTPS	If you need to retrieve data via HTTPS, you can submit the command `set odps.sql.unstructured.data.oss.use.https=true;` along with the SQL statement.	2020-02-26	All regions	ORC external table
External tables support the `MSCK REPAIR TABLE` syntax.	MaxCompute can automatically add partitions to an external table based on the file paths of its data in OSS.	2020-02-26	All regions	ORC external table

January

Feature	Description	Release date	Region	References
Support for DATE and CHAR data types	You can enable data type version 2.0 to use the DATE and CHAR data types when you create or read tables in MaxCompute.	2020-01-03	All regions	Data type version guide

2019

December

Feature	Description	Release date	Region	Related documentation
Support for open source geospatial UDFs	You can use open source geospatial UDFs to analyze spatial data.	2019-12-25	All regions	UDFs (geospatial)

November

Feature	Description	Release date	Region	Related documentation
Support for connecting to MaxCompute with Tableau	You can use the MaxCompute JDBC driver to connect Tableau to MaxCompute and perform visual analysis on your data.	2019-11-18	All regions	Configure JDBC to use Tableau
New MaxCompute JDBC driver	The MaxCompute JDBC driver is a Java Database Connectivity (JDBC) interface that allows you to run distributed queries on massive data.	2019-11-18	All regions	JDBC overview

October

Feature	Description	Release date	Region	Related documentation
Support for adding or dropping multiple partitions at once	MaxCompute now supports adding or dropping multiple partitions in a single DDL operation, which simplifies batch partition management and improves data warehousing efficiency.	2019-10-28	All regions	Announcement
Support for new operators	MaxCompute now supports the`IS [NOT] DISTINCT FROM` operator and the`\|\|` string concatenation operator.	2019-10-28	All regions	Announcement
Support for NATURAL JOIN	MaxCompute now supports NATURAL JOIN to expand its SQL JOIN capabilities.	2019-10-28	All regions	JOIN
Support for specifying default column values	You can specify default values for columns when you create a table.	2019-10-28	All regions	Announcement
Support for LIMIT OFFSET	MaxCompute now supports LIMIT OFFSET to enhance its SQL engine capabilities.	2019-10-28	All regions	SELECT syntax
Support for odps.sql.orderby.position.alias and odps.sql.groupby.position.alias	MaxCompute now supports odps.sql.orderby.position.alias and odps.sql.groupby.position.alias. These parameters allow integer constants in ORDER BY and GROUP BY clauses to be treated as column sequence numbers in the SELECT list.	2019-10-28	All regions	SELECT syntax
New built-in functions: `JSON_TUPLE()` and `EXTRACT()`	The `JSON_TUPLE()` and `EXTRACT()` built-in functions are now available to enhance SQL processing.	2019-10-28	All regions	String functions Date and time functions
Support for specifying JOIN priority with parentheses	You can use parentheses to specify the priority of JOIN operations.	2019-10-28	All regions	JOIN
Support for merging partitions	If you have a large number of partitions, you can use the partition merging feature to archive data.	2019-10-28	All regions	Partition operations

September

Feature	Description	Release date	Region	Related documentation
Support for Information Schema	MaxCompute's Information Schema provides information such as project metadata and usage history data.	2019-09-06	All regions	Information Schema overview
Storage price reduction	MaxCompute reduced its storage pricing on September 1, 2019. The previous five pricing tiers were adjusted to three, and the unit price for each tier was lowered.	2019-09-01	All regions	None.

July

Feature	Description	Release date	Region	Related documentation
Support for Spark service	Spark on MaxCompute is a service compatible with open source Spark. It runs a Spark computing framework on MaxCompute's unified resource and permission system. This allows you to run Spark jobs in a familiar environment to address diverse data processing needs.	2019-07-23	All regions	Overview
New region: China (Chengdu)	MaxCompute is now available in the China (Chengdu) region.	2019-07-04	China (Chengdu)	None.

June

Feature	Description	Release date	Region	Related documentation
Support for obtaining ExecutionContext objects in UDTs	UDTs can obtain an ExecutionContext object to access information about their runtime environment.	2019-06-25	All regions	UDT overview
Support for function overloading in UDTs	UDTs support generics, class inheritance, and variable-length parameters to provide flexible function definitions.	2019-06-25	All regions	UDT overview
Support for dynamic parameters in UDAFs and UDTFs	MaxCompute now supports dynamic parameters for UDAFs and UDTFs and extends the Resolve annotation syntax, making UDF development more flexible.	2019-06-24	All regions	Dynamic parameters of UDAFs and UDTFs
Support for partition pruning in UDFs	UDFs support partition pruning to reduce computing costs.	2019-06-24	All regions	None.
Parameterized views	Parameterized views support passing any table or other variables to customize the view.	2019-06-24	All regions	Parameterized views
Support for generating CREATE TABLE DDL	You can use the `SHOW CREATE TABLE` statement to generate the DDL for creating a table. This simplifies recreating table schemas.	2019-06-24	All regions	Table operations
Support for configuring monitoring and alerts for subscription job spending	You can use CloudMonitor to monitor spending on subscription jobs and configure alerts.	2019-06-21	All regions	Monitoring and alerts