MaxCompute:2025

December 4, 2025 - Additional CLONE TABLE usage limits

• The CLONE TABLE command cannot be used to copy table data between projects that have different storage types. For example, you cannot use the CLONE TABLE command to copy a table from a project that uses multi-zone storage to a project that uses single-zone storage, and vice versa.

• Data replicas are distributed differently in multi-zone and single-zone storage projects. Therefore, you cannot use CLONE TABLE to quickly copy a full table by cloning only its metadata. You must wait for the data to be completely rewritten across availability zones (AZs) in the background.

• Recommended solutions for copying tables between projects that have different storage types:

  • For non-partitioned tables, use CREATE TABLE AS to migrate data.

  • For partitioned tables, first create a table by using CREATE TABLE LIKE, and then run INSERT OVERWRITE to migrate data.

If you have any questions or need technical support, contact us or submit a ticket.

December 1, 2025 - MaxCompute network upgrade impact

From December 1, 2025 (Monday) to December 31, 2025 (Wednesday), China Standard Time (UTC+8), MaxCompute will gradually upgrade the leased line network in all regions.

This upgrade affects the following:

• Network connections for users who have revoked permissions from or modified the permission policy of the aliyunodpsroleforeni role.

• Network connections for users with an overdue payment.

Solutions:

• If you have revoked permissions from or modified the permission policy of the aliyunodpsroleforeni role, you must log on with the primary account that was used to create the network connection and click Authorize to grant the permissions again. For more information, see the network connection process for MaxCompute.

• If your account has an overdue payment, pay the outstanding balance to resolve the issue.

If you do not resolve these issues by December 31, affected jobs may fail.

If you have any questions or need technical support during or after the upgrade, contact us or submit a ticket.

November 5, 2025 - Ubuntu operating system support

MaxCompute image management now supports the Ubuntu operating system ecosystem. This provides developers with improved compatibility and scalability to build more diverse computing environments. This update covers built-in images and custom images. The technical details are as follows:

• Built-in images

  A new ubuntu base operating system image, built on the Ubuntu 20.04 LTS architecture, is now available. It is compatible with the Python 3.7 and Python 3.11 runtime environments. You can seamlessly use it for MaxCompute SQL user-defined function (UDF) development and integration with the MaxFrame distributed AI computing engine framework. This significantly improves cross-platform code migration efficiency and execution performance.

• Custom images

  An Ubuntu base image is now available, which you can use with a Dockerfile to build custom images. The address of the MaxCompute Ubuntu base image is registry.cn-zhangjiakou.aliyuncs.com/maxcompute_image/ubuntu_20.04:latest.

November 4, 2025 - MaxCompute Logview security upgrade

Starting from December 17, 2025 (Wednesday), China Standard Time (UTC+8), MaxCompute will progressively roll out a comprehensive security upgrade for Logview in the China (Beijing), China (Hangzhou), China (Shanghai), China (Hangzhou) Finance, China (Shanghai) Finance, China (Shenzhen) Finance, and China (Beijing) Gov regions.

After the upgrade, Logview links for jobs submitted from the shared resource group for scheduling in DataWorks (Data Development and Data Analysis) will redirect to the MaxCompute console. Additionally, the upgraded Logview will not display job results by default.

If you have any questions about Logview access or permission configuration, or need technical support during or after the upgrade, contact us or submit a ticket.

October 30, 2025 - New SQL analysis features

To adapt to the evolution of cloud-native technologies and meet the increasingly complex demands of data analysis, the SQL analysis feature in the MaxCompute console has been upgraded. It is deeply integrated with the new version of Data Development (DataStudio) in DataWorks to provide a more efficient, intelligent, and unified experience for data analysis and metadata management. In addition to built-in data from MaxCompute public datasets and SQL demo scripts for public datasets, the new SQL analysis feature also supports capabilities such as an intelligent SQL editor and a data catalog.

This feature is currently available in the Indonesia (Jakarta) region, with more regions to be supported in November. Before you use this feature, take note of the following:

• To use the new SQL analysis feature, you must activate the pay-as-you-go (free) DataWorks Basic Edition or any other edition.

• The SQL file management method has changed. The new SQL analysis feature provides a personal folder for you to create and manage SQL directories and files in the cloud. Existing local files from the old workspace are not automatically synchronized to the personal folder. We recommend manually moving these files to the personal folder for future editing and use.

For more information about the features and how to enable them, see Connect by using the new SQL analysis feature.

If you encounter any issues while using the new SQL analysis feature, contact us or submit a ticket.

October 29, 2025 - Append Delta Table is now supported in all regions

Delta Table is a high-performance table format developed by MaxCompute for large-scale analytical datasets. A Delta Table can be an Append Delta Table without a primary key or a PK Delta Table with a primary key.

This feature is now supported in all regions.

September 19, 2025 - Updates for Hudi and Delta Lake external tables discontinued

• The default Hudi SDK version integrated into the MaxCompute Hudi external table system is org.apache.hudi:hudi-hadoop-mr-bundle:0.12.2-emr-1.0.6. Starting from September 19, 2025, MaxCompute will support only the existing external table features integrated with this Hudi SDK version, and no further SDK updates or feature enhancements will be provided.

• The default Delta Lake SDK version integrated into the MaxCompute Delta Lake external table system is io.delta:delta-core_2.11:0.2.0.5. Starting from September 19, 2025, MaxCompute will support only the existing external table features integrated with this Delta Lake SDK version, and no further SDK updates or feature enhancements will be provided.

If you require more capabilities for these two types of external tables, we recommend that you use a Paimon external table as an alternative.

September 16, 2025 - Append Delta Table released

Delta Table now supports Append Delta Table for tables without primary keys. The original Delta Table, which only supported tables with primary keys, is now named PK Delta Table.

Delta Table is a high-performance table format developed by MaxCompute for large-scale analytical datasets. A Delta Table can be an Append Delta Table without a primary key or a PK Delta Table with a primary key. It supports features such as ACID transactions, incremental queries, time travel queries, dynamic cluster bucketing, real-time data updates, and schema evolution.

The native lakehouse and near-real-time computing capabilities of the MaxCompute platform allow you to use standard SQL to create, update, and query a Delta Table. You do not need to manage the complex underlying storage and metadata, as MaxCompute automatically maintains and optimizes them. This provides both ease of use and cost-effectiveness.

The new Delta Table format provides an improved user experience:

• Balances data freshness and query efficiency for real-time services.

• Adapts to business growth by overcoming the limitations of previous table formats.

• Balances performance and reliability.

September 10, 2025 - Multiple level-2 quotas for data transfer resource groups

Starting from 19:00 on September 17, 2025 (Wednesday), China Standard Time (UTC+8), MaxCompute will upgrade the quota management for exclusive resource groups for data transfer. After the upgrade:

• The quota configuration for MaxCompute data transfer services will be fully unified with the quota configuration for computing resources. This allows you to manage multiple resource types with a single set of rules.

• You can create multiple level-2 quotas under a level-1 quota for the MaxCompute data transfer service. This allows you to share purchased resources across businesses while ensuring resource isolation. You can also flexibly set configuration plans and schedules based on business cycles, such as peak and off-peak hours, to avoid wasting resources and reduce costs.

The pre-upgrade version does not support setting configuration plans and schedules. If you are already using the time-sharing feature for the data transfer service, the system generates a corresponding configuration plan to ensure that your services are not affected by this upgrade. You can go to the console to modify the configuration plan.

August 22, 2025 - Console menu bar optimization

Starting from 21:00 on September 11, 2025 (Thursday), China Standard Time (UTC+8), MaxCompute will upgrade its console menu bar. The upgrade improves the user experience in the following ways:

• Quickly locate target features or content by using the search function in the menu bar. This reduces clicks, simplifies the interface, enhances usability, and optimizes the information retrieval experience.

• The left-side navigation pane improves search efficiency with clear categorization and logical arrangement. Frequently used features are expanded by default, which reduces clicks and improves operational accessibility and smoothness.

• Infrequently used features are grouped under the More option. Tenant Property and Disaster Recovery are moved to submenus, with related features consolidated into tabs.

The following table maps the menu items from the old version to the new version.

July 25, 2025 - Reverse VPC access for Spark deprecated

We are ending support for the reverse VPC access solution for Spark due to a technology upgrade. MaxCompute plans to deprecate this feature on September 30, 2025 (Tuesday), China Standard Time (UTC+8). After this date, you cannot use this feature for new jobs.

We recommend that you upgrade your existing MaxCompute Spark reverse VPC access settings to the direct connection over leased line method as soon as possible. Follow these steps to upgrade:

1. Enable the leased line network connection.

   For more information about the process, see VPC access solution (direct connection over leased line).

2. Modify the Spark job's configuration to use the direct connection over leased line method.

   • Previously, if you used the reverse VPC access method, your Spark job was configured with the spark.hadoop.odps.cupid.vpc.domain.list parameter. The following code provides a parameter example:

     Parameter example

     {
       "regionId":"cn-beijing", 
       "vpcs":[
         {
           "vpcId":"vpc-2zeaeq21mb1dmkqh*****",
           "zones":[
             {
               "urls":[
                 {
                   "domain":"hb-2zecxg2ltnpeg8me4-master*-***.hbase.rds.aliyuncs.com",
                   "port":2181
                 }
               ]
             }
           ]
         }
       ]
     }
     

   • To switch to the direct connection over leased line method, add the following two configuration parameters to your Spark job. You need to replace RegionID and VPCID with their actual values.

     # Add the following configurations to the spark-defaults.conf file in DataWorks.
     spark.hadoop.odps.cupid.eni.enable = true
     spark.hadoop.odps.cupid.eni.info = <regionid>:<vpcid>

     Example:

     # For the preceding reverse VPC access example, add the following configuration parameters.
     spark.hadoop.odps.cupid.eni.enable = true
     spark.hadoop.odps.cupid.eni.info = cn-beijing:vpc-2zeaeq21mb1dmkqh*****

     After the configuration is complete, you can view the JSON summary of the job on the Json Summary tab of Logview. Confirm that the vpc_mode field is set to eni, which indicates the ENI direct connection method is used.

     {
         "Inputs" :
         {
     
         },
         "StorageTierInputs" :
         {
     
         },
         "cost_cpu" : "122400",
         "cost_indicators" :
         {
             "master" :
             {
                 "cpu_plan" : 200,
                 "cpu_usage_avg" : 51.6666666666667,
                 "mem_plan" : 9011,
                 "mem_usage_max" : 719
             },
             "worker" :
             [
     
             ]
         },
         "cost_memory" : "3081762",
         "end_time" : "1753069636",
         "start_time" : "1753069564",
         "vpc_mode" : "eni"
     }
     

July 24, 2025 - MaxCompute network connection solution upgrade

From July 14, 2025 (Monday) to July 31, 2025 (Thursday), China Standard Time (UTC+8), MaxCompute will upgrade the underlying technical solution for the network connection feature in China public cloud regions. The details are as follows:

• Before the upgrade (original solution): Two elastic network interfaces (ENIs) are created in your VPC.

• After the upgrade (new solution): The system automatically creates two reverse endpoints (PVLs) and two security groups to manage the PVL security rules, which implements the MaxCompute network connection.

After the new network solution is launched, you can see two PVLs on the Endpoint page of the VPC console and two security groups on the Network & Security > Security Group page of the ECS console. You will also receive an SMS notification that PrivateLink is enabled.

This upgrade reinforces the network architecture and will not affect your business or fees.

The two security groups are prefixed with MaxCompute (for example, MaxCompute-vpc-xxx and MaxCompute-backup-vpc). The Description column shows Create by MaxCompute, and the security group type is Basic Security Group.

For more feature information, see Network connection process.

March 24, 2025 - Multi-zone storage specification launched

MaxCompute has officially launched the multi-zone storage specification. The original default storage specification is renamed to single-zone storage. For more information about storage specifications and billing, see Storage fees.

We recommend selecting multi-zone storage for enterprise production data to protect against AZ-level failures. In the event of an AZ-level failure, multi-zone storage ensures that data read and write services are not interrupted and prevents data loss. For more information about how to use multi-zone storage, see Zone disaster recovery.

After the new storage specification is launched, MaxCompute will also optimize the naming of billable items in your bills. The naming adjustments will not affect your billing or business operations.

If you have any questions or need assistance, contact us through your dedicated DingTalk group or submit a ticket.

March 18, 2025 - Clustering optimization recommendation released

On March 17, 2025 (Monday), China Standard Time (UTC+8), MaxCompute released the clustering optimization recommendation feature in China public cloud regions. This feature analyzes the recent read and write patterns of tables and generates corresponding clustering suggestions to help you improve job performance and reduce CU consumption. You can decide whether to apply the system-recommended clustering suggestions based on the estimated benefits and recommendation details. You can also view the daily benefits after applying the suggestions. For more information, see Clustering optimization recommendation.

If you have any questions or need assistance while using this feature, contact us through your dedicated DingTalk group or submit a ticket.

March 18, 2025 - Cost estimation for custom policies in tiered storage optimization

On March 14, 2025 (Friday), China Standard Time (UTC+8), MaxCompute released a feature that allows you to estimate storage cost changes based on custom policies in addition to the default policy. This helps you decide how to configure tiered storage policy attributes. For more information, see Cost estimation for custom policies.

If you have any questions or need assistance while using this feature, contact us through your dedicated DingTalk group or submit a ticket.

March 17, 2025 - Price reduction for disaster recovery storage

Effective 00:00 on March 31, 2025 (Monday), China Standard Time (UTC+8), MaxCompute will reduce the unit price of pay-as-you-go storage for disaster recovery by approximately 25%. This price reduction improves data storage disaster recovery capabilities and meets the demand for lower-cost multi-zone storage. The specific adjustments are as follows:

This price adjustment will not affect your business or operations.

If you have any questions or need assistance, contact us through your dedicated DingTalk group or submit a ticket.

February 21, 2025 - Automatic materialized view (AutoMV) released

Starting from February 20, 2025 (Thursday), China Standard Time (UTC+8), MaxCompute will gradually release the automatic materialized view (AutoMV) feature in China public cloud regions. You can enable this feature in the MaxCompute console. After you enable the feature, it is expected to take effect the next day. MaxCompute will then automatically analyze the project's job data in the background and create, manage, and apply materialized views to save computing costs and accelerate job execution. Typically, on the third day after you manually enable the feature, you can view the actual benefits and storage usage in the MaxCompute console. For more information, see Automatic materialized view (AutoMV).

During this process, user jobs are almost unaffected. You need to pay for the storage of the system-created materialized views at the standard storage unit price. No additional service fees are incurred.

To help you better understand and use the MaxCompute intelligent optimization feature, the console menu bar has been adjusted as follows:

If you have any questions or need assistance while using this feature, contact us through your dedicated DingTalk group or submit a ticket.

February 13, 2025 - Public preview of the MaxQA query acceleration engine

Starting from 09:00 on February 17, 2025 (Monday), China Standard Time (UTC+8), MaxCompute will gradually begin the public preview of the new MaxQA query acceleration engine (MaxQA) for projects in some regions. The details are as follows:

MaxQA is designed to optimize query performance in near-real-time scenarios. It optimizes multiple stages including the control link, query optimizer, execution engine, storage engine, and caching mechanism within a dedicated query computing resource pool. It is suitable for low-latency and high-stability business scenarios such as BI, interactive analysis, and near-real-time data warehouses. The dedicated computing resources required by MaxQA are a level-2 quota of the interactive type under a subscription level-1 quota resource group. For usage details, see MaxQA basic operations.

During the public preview period, you must submit a ticket to enable the MaxQA feature. To help you use this feature, we provide a one-month trial of 100 CU of computing resources (valued at USD 2,200) to reduce your trial costs.

If you have any questions or need assistance while using this feature, contact us through your dedicated DingTalk group or submit a ticket.

January 15, 2025 - Console Overview page released

Starting from 17:00 on January 16, 2025 (Thursday), China Standard Time (UTC+8), MaxCompute will gradually release the Overview page in all regions.

The Overview page provides a quick start guide, a list of global resource instances, lists of favorite projects and quotas with quick actions, alerts and risk warnings, and product updates. When you log on to the MaxCompute console, you are directed to the Overview page by default to help you get started with your O&M and management tasks.

If you have any questions or need assistance while using this feature, contact us through your dedicated DingTalk group or submit a ticket.