MaxCompute:2024

December 23, 2024: MaxFrame adds support for new regions

Dear MaxCompute users,

MaxCompute's distributed computing framework, MaxFrame, will become available in the Singapore, Japan (Tokyo), and Indonesia (Jakarta) regions starting Monday, December 23, 2024 (Beijing time, UTC+8).

MaxFrame is a distributed computing framework developed in-house by Alibaba Cloud. It supports a Python programming interface and directly uses the computing resources and data interfaces of MaxCompute. MaxFrame is compatible with Pandas and XGBoost interfaces and automatically handles distributed processing, allowing Python developers to efficiently perform large-scale data processing, ML/AI data preprocessing, and model development on MaxCompute. For more information, see MaxFrame: The distributed AI compute engine.

For questions or assistance, contact us in the MaxCompute Developer Community (DingTalk group ID: 11782920), your dedicated DingTalk group, or the official MaxFrame user support group (DingTalk group ID: 37130012987).

Thank you.

November 28, 2024: MaxCompute Service Level Agreement update

Dear MaxCompute users,

Alibaba Cloud is updating the service level agreement (SLA) for MaxCompute, our big data computing service. The main changes are as follows:

• Addition of clauses defining service unavailability in Sections 1.1 to 1.10.

• Addition of exception clauses (k), (l), and (m) to Section 1.11.

For complete details, see the MaxCompute Service Level Agreement.

The new service level agreement (SLA) will take effect on Wednesday, January 1, 2025 (UTC+8). If you have any questions, contact us through your dedicated DingTalk group or the MaxCompute Developer Community group (DingTalk group ID: 11782920).

Thank you for your continued support of Alibaba Cloud.

November 14, 2024: MaxCompute LogView security upgrade

Dear MaxCompute users,

MaxCompute is rolling out a comprehensive security upgrade for LogView, starting at 14:00 on Monday, December 2, 2024 (UTC+8). This upgrade will be gradually rolled out to the following regions: China (Hong Kong), China (Shanghai) Finance, China (Beijing) Finance (invitation-only preview), China (Beijing) Gov 1, China (Shenzhen) Finance, Japan (Tokyo), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Germany (Frankfurt), UK (London), US (Silicon Valley), US (Virginia), UAE (Dubai), and SAU (Riyadh - Partner Region). After the upgrade, LogView links for jobs submitted from DataWorks DataStudio or DataAnalysis using the public scheduling resource group will redirect to the MaxCompute console. To access the LogView for a job, you must meet the following requirements:

• Have a valid Alibaba Cloud account with real-name verification.

• Be a member of the MaxCompute project where the job runs. A project administrator can add you to the project.

Therefore, be aware that if the access identity configured for the MaxCompute data source in your DataWorks workspace differs from your user identity, you may not be able to access the LogView for a job you submitted. This is because you are not a member of the MaxCompute project that owns the job. For example, in a standard DataWorks workspace, if you submit a job from the development environment to the production environment, you cannot access its LogView if you are not a member of the production environment's MaxCompute project.

To prevent this issue, ensure the following:

• All team members who need to access LogView must have an Alibaba Cloud account with real-name verification.

• Promptly add team members to the relevant MaxCompute projects and grant them the required permissions on specific objects.

  Note

  For more information about user management and authorization, see User planning and management and MaxCompute permissions.

Additionally, after the upgrade, LogView will not display job execution results by default.

During or after the upgrade, if you encounter any issues with LogView access or permission configuration, or require technical support, contact us through your dedicated DingTalk group or the MaxCompute Developer Community group (DingTalk group ID: 11782920).

Thank you for your cooperation!

November 13, 2024: MaxCompute intelligent job diagnostics

Dear MaxCompute users,

MaxCompute is launching intelligent diagnostics for jobs, with a phased rollout in all regions starting Wednesday, November 13, 2024 (UTC+8). This feature analyzes the execution details of SQL jobs to:

• For failed jobs: It identifies the cause of failure based on the error message and provides a detailed description with a recommended solution.

• For slow-running jobs: It analyzes resource usage and execution details to diagnose exceptions such as resource insufficiency, data skew, data bloat, and mode rollback, providing the root cause and solution for each.

This feature helps you quickly identify and resolve issues such as job failures or unexpected runtimes. For more information and usage instructions, see Intelligent Diagnostics for Jobs.

For questions or assistance, contact us through your dedicated DingTalk group or the MaxCompute developer community (DingTalk group ID: 11782920).

Thank you.

October 30, 2024: MaxCompute tunnel upgrade and cross-region traffic governance

Starting Monday, November 18, 2024 (UTC+8), MaxCompute will begin a phased upgrade of its tunnel service and implement cross-region traffic governance. The key changes include:

• Improved tunnel performance and traffic control

  • For Streaming Tunnel, the write throughput limit for a single slot is increased from 1 MB/s to 10 MB/s, and the request limit is raised from 10 to 100 requests per second per slot. For Batch Tunnel, the write throughput limit for a single slot is now 10 MB/s.

  • The slot usage calculation for Streaming Tunnel is optimized. Instead of reserving a specified number of slots when you create a StreamUploadSession, one slot is now consumed when a RecordPack is written and released immediately upon completion, which aligns with the calculation method used for Batch Tunnel. For more information, see Tunnel overview.

• Cross-region traffic governance

  To enhance security, MaxCompute will prohibit connections to a MaxCompute project from an endpoint outside the project's region. You must use the endpoint or Tunnel endpoint specific to your project's region. For a list of available endpoints, see Endpoints.

If you have any questions or need assistance, contact us through your dedicated DingTalk group or the MaxCompute Developer Community DingTalk group (ID: 11782920).

Thank you for your cooperation!

October 28, 2024: Behavioral changes to UPDATE/DELETE/MERGE INTO from partition conversion

Dear MaxCompute users,

As of Thursday, October 24, 2024 (UTC+8), MaxCompute has enabled the dynamic-to-static partition conversion optimization by default for all DML operations to improve query performance. This optimization affects the behavior of UPDATE/DELETE/MERGE INTO operations. If you run one of these operations on a partition that does not exist, the operation will now fail and return an error similar to the following:

• ODPS-0110061: Failed to run ddltask - Operation DELETE expect target partition exist, but object not found.

• ODPS-0422141: Partition not found - Partition 'projectName.tableName(pt=xxx)' does not exist.

If this issue occurs, you can disable this optimization at the session level by running the SET odps.optimizer.cbo.rule.filter.black=re; command. You can then run UPDATE/DELETE/MERGE INTO operations as before.

For questions or assistance, contact us through your private DingTalk group or the MaxCompute developer community group (DingTalk group ID: 11782920).

Thank you for your cooperation!

October 22, 2024: MaxCompute SDK for Go released

Dear MaxCompute users,

The MaxCompute SDK for Go is now generally available, effective Tuesday, October 22, 2024 (UTC+8). The SDK for Go provides APIs to operate and manage MaxCompute services, such as accessing and managing projects, working with data tables, and performing data transfers. For more information, see Overview of the SDK for Go.

If you have any questions or need assistance, contact us through your dedicated DingTalk group or the MaxCompute developer community group (DingTalk group ID: 11782920).

Thank you for your support.

October 15, 2024: MaxCompute classic network retirement and endpoint updates

Dear MaxCompute users,

Alibaba Cloud will retire the classic network connection for MaxCompute on Saturday, November 30, 2024 (UTC+8) and replace it with the internal network for connecting cloud services. To prevent service disruption, please switch to a VPC or the internal network for connecting cloud services before this date. Additionally, we have updated the endpoints for VPC and internal network connections in Alibaba Finance Cloud, Alibaba Gov Cloud, and other international regions. Please review and update your configurations accordingly. For details, see Endpoints.

If you have any questions or need assistance, contact us through your dedicated DingTalk group or the MaxCompute Developer Community group (DingTalk group ID: 11782920).

Thank you for your cooperation.

October 14, 2024: Support for multiple subscription resources

Dear MaxCompute users,

Starting Monday, October 14, 2024 (UTC+8), MaxCompute will support purchasing multiple subscription computing resources for a single Alibaba Cloud account in the same region. This means you can now purchase multiple level-1 quotas, and the resources for each quota are isolated from one another.

This feature is ideal for the following use cases:

• Your business requires strong isolation for subscription computing resources between different departments or projects within a single Alibaba Cloud account.

• Your business needs to split billing for different departments or projects. Previously, you could not use a level-2 quota to separate charges for subscription computing resources. You can now purchase multiple level-1 quotas to split your bills by order.

For more information about MaxCompute quotas, see Manage computing resources and quotas.

For questions or assistance, contact us through your dedicated DingTalk group or the MaxCompute developer community group (DingTalk group ID: 11782920).

Thank you.

September 27, 2024: Storage cost optimization for MaxCompute

Dear MaxCompute users,

Starting Friday, September 27, 2024 (UTC+8), MaxCompute is rolling out the storage cost optimization feature in public cloud regions in the Chinese mainland. This feature helps you optimize your tiered storage policy by analyzing the last access time of your tables and partitions. Based on a fixed policy, it recommends moving data from standard storage to lower-cost tiers: to infrequent access (IA) after 30 days of inactivity, and to long-term storage after 180 days of inactivity. The feature also estimates your potential cost savings based on the catalog price to help you make an informed decision.

For more information, see Optimize tiered storage configurations.

For questions or assistance, contact us through your dedicated DingTalk group or the MaxCompute developer community (DingTalk group ID: 11782920).

Thank you.

September 26, 2024: General availability of MaxFrame for MaxCompute

Dear MaxCompute users,

MaxFrame is a distributed computing framework from Alibaba Cloud that uses MaxCompute's compute resources and data interfaces. It provides a Python interface, is compatible with Pandas and XGBoost, and automatically handles distributed processing. With MaxFrame, Python developers can efficiently perform large-scale data processing and ML/AI data preprocessing on MaxCompute. For more information: MaxFrame: the distributed AI compute engine.

Starting Friday, September 27, 2024 (UTC+8), MaxFrame is generally available in the China (Beijing), China (Hangzhou), China (Shanghai), China (Shenzhen), China (Chengdu), China (Ulanqab), and China (Hong Kong) regions. MaxFrame is billed based on the compute resources used by jobs. Both pay-as-you-go and subscription billing models are supported. For billing details, see Compute costs (subscription).

For questions or assistance, contact us through the MaxCompute Developer Community (DingTalk group ID: 11782920), your dedicated DingTalk group, or the official MaxFrame user support group (DingTalk group ID: 37130012987). To join the MaxCompute Developer Community, fill out the DingTalk group application form.

Thank you for your continued support.

September 20, 2024: RAM permission management for MaxCompute console features

Dear MaxCompute users,

To enhance security, MaxCompute will introduce RAM permission management for several console features on Tuesday, October 8, 2024. While an Alibaba Cloud account (the primary account) retains full access by default, RAM users will require specific permission actions to use these affected features. The features and their required permission actions are listed below:

For details on these permission actions, see RAM permissions.

To ensure uninterrupted access to these console features, grant the required RAM permissions to affected RAM users before Tuesday, October 8, 2024.

If you have any questions or need assistance, contact us through your dedicated DingTalk group or the MaxCompute developer community group (DingTalk group ID: 11782920).

Thank you for your cooperation.

September 19, 2024: MaxCompute new features invitational preview

Dear MaxCompute users,

We are excited to announce an invitational preview of major new features of MaxCompute, launched at the 2024 Apsara Conference. These new capabilities include data lakehouse 2.0 powered by OpenLake, processing unstructured data in object tables using SQL or MaxFrame, the incremental Delta Table format, incremental computing based on incremental materialized views, and enhanced query performance with the MaxCompute Query Acceleration (MCQA) 2.0 engine. The preview is available in the China (Hangzhou), China (Shanghai), China (Beijing), and China (Shenzhen) public cloud regions. To explore these new features, please complete the invitational preview application form.

After you register, you can explore the new features in one of the following ways:

• In the MaxCompute console, when you create a new project, set the Project Type to Project for Invitational Preview of New Features at Apsara Conference.

• In the DataWorks Data Studio public preview, when you bind MaxCompute computing resources and create a new internal project, set the Project Type to Project for Invitational Preview of New Features at Apsara Conference.

September 5, 2024: MaxCompute quota management UI enhancements

Dear MaxCompute users,

On Wednesday, September 11, 2024, MaxCompute will progressively roll out an optimized version of the console UI for configuring subscription compute quotas to all regions. This version is optimized for two main use cases: with and without the need for time-based scaling. If you do not need to use the time-based scaling feature, you can complete the configuration of your subscription compute quota on the Basic Configurations tab in the new UI. If you need to use the time-based scaling feature, you can first complete the Basic Configurations and then go to the Scaling Configuration tab to make the relevant settings.

When you configure subscription compute Quotas on the Quotas page in the MaxCompute console, the following changes will apply:

• The original quota template has been renamed to Basic Configurations. The capabilities are as follows:

  • You can use Edit Basic Configurations to add a level-2 quota and modify its basic parameters, such as Exclusive, CU Limit per Job, Enable Priority, Non-reserved CUs, and Scheduling Policy. You can also edit the Reserved CUs [minCU,maxCU] and Elastically Reserved CUs. However, these changes apply only to the currently active quota resource configuration plan (formerly quota plan). If you use time-based scaling, we recommend that you go to the Scaling Configuration tab to configure the Reserved CUs [minCU,maxCU] and Elastically Reserved CUs.

  • On the Basic Configurations tab, you can delete level-2 quotas from the Quota list. The deletion logic is consistent with that in the original Quota template.

• The original Quota Plans and Time Plan are merged into Scaling Configuration, and configuration is limited to Reserved CUs [minCU,maxCU] and Elastically Reserved CUs. The capabilities are as follows:

  • Add configuration plan: This action was formerly Add Plan and supports configuring only the Reserved CUs [minCU,maxCU] and elastically reserved CUs parameters for a quota.

  • Resource Configuration Plan List: This list, formerly the quota plan list, can now be expanded to view and compare the configuration details of each plan.

  • Time-based Management: This feature was previously called Time Plan. The functionality remains the same.

  • Time-based Scaling Preview: This is a new feature that displays a trend chart of CU allocation over different time periods for a selected level-2 quota.

  • Global Scaling Configuration View: This is a new feature that provides a table of all configuration plan details. This allows you to easily check the Min/Max CUs and elastically reserved CUs for all quotas across different time slots.

For detailed instructions, see Computing resources - Quota management. If you have any questions or need assistance, contact us through your dedicated DingTalk group, the official MaxFrame user support group (DingTalk group ID: 37130012987), or the MaxCompute Developer Community group (DingTalk group ID: 11782920).

Thank you for your cooperation!

August 29, 2024: RAM permission control for MaxCompute console features

Dear MaxCompute users,

To enhance security, MaxCompute will add RAM permission management to specific console features starting on Tuesday, September 10, 2024. By default, an Alibaba Cloud account (primary account) has full access to all console features. However, RAM users will require the associated permission actions to access these features. The affected features and their permission actions are listed below:

For details on these permission actions, see RAM permissions.

To ensure continued access for your RAM users, grant them the required RAM permissions before Tuesday, September 10, 2024. For example, to grant a RAM user permission to manage quotas, attach the following policy:

{
  "Version": "1",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": [
        "odps:UpdateQuota",
        "odps:UpdateQuotaPlan",
        "odps:UpdateSubQuotas",
        "odps:UpdateQuotaSchedule",
        "odps:CreateQuotaPlan",
        "odps:DeleteQuotaPlan",
        "odps:CreateQuotaSchedule",
        "odps:ListQuotaRoutingRules",
        "odps:CreateQuotaRoutingRule",
        "odps:GetQuotaRoutingRule",
        "odps:RemoveQuotaRoutingRule",
        "odps:UpdateQuotaRoutingRule"         
      ],
      "Resource": "*"
    }
  ]
}

For questions or assistance, contact us through your dedicated DingTalk group or the MaxCompute developer community group (DingTalk group ID: 11782920).

Thank you for your cooperation!

August 19, 2024: Enhancements to MaxCompute resource observation

Dear MaxCompute users,

The MaxCompute resource observation feature now includes new metrics for storage resources and job performance.

• Storage resources

  This feature helps you track changes in your storage usage and understand its distribution. You can view the total storage usage and the proportion of different storage types in a selected region. You can also observe storage trends for specific projects over a chosen time period and view detailed information for each table or partition. With these insights, you can identify any sudden spikes in storage consumption and promptly optimize the top-consuming projects, tables, or partitions.

• New metrics for job performance

  • Identify the top 10 jobs that consume the most CPU-hours or memory-hours within your selected filters. You can also view the top 10 signatures and ExtNodeIds with the highest total or average consumption.

  • View the top 10 jobs with the longest total runtimes, along with the top 10 signatures and ExtNodeIds with the longest total or average runtimes, based on your selected filters.

  • View the trend of data scanned by jobs over a specific period.

  • Track the trend of data scanned by jobs per CU-hour.

  These metrics help you quickly pinpoint resource-intensive and long-running jobs for timely optimization. By comparing the data scanned by jobs per CU-hour, you can detect periods of abnormal performance. You can then use the job management feature to investigate the root cause and resolve the issue.

For more information, see Resource observation.

If you have any questions, contact us via your dedicated DingTalk group or the MaxCompute Developer Community DingTalk group (ID: 11782920).

Thank you.

July 31, 2024: Retirement of legacy PyTorch components

Dear MaxCompute users,

Due to a platform-wide system upgrade, Platform for AI (PAI) will retire legacy PyTorch algorithm components across all clusters, effective August 30, 2024. We strongly recommend that you migrate any PyTorch jobs submitted to MaxCompute using the PAI command pai -name pytorch100/pytorch131/pytorch151. We recommend using Deep Learning Containers (DLC) to submit PyTorch jobs. For more information, see Create a training job. Starting August 31, 2024, our service-level agreement (SLA) will no longer cover existing jobs that use these retired components.

If you have any questions or require assistance, contact us through your dedicated DingTalk group or the MaxCompute developer community (DingTalk group ID: 11782920).

Thank you for your cooperation.

July 18, 2024: Retirement of specific APIs

Dear MaxCompute users,

Effective Wednesday, July 31, 2024 (UTC+8), MaxCompute will retire the following api operations:

• Retired api operations and their replacements

  Retired API		Replacement API
  Version	API name
  2019-06-12	CreateProject	2022-01-04: CreateProject
  	DeleteProject	Not available
  	GetProject	2022-01-04: GetProject
  	OpenMaxComputeService	Not available
  2022-01-04	GetQuota	2022-01-04: QueryQuota

• Reasons for retirement

  • We are retiring the api operations in version 2019-06-12 because the underlying feature implementation has changed, and long-term monitoring shows no external customer calls to this version.

  • We are updating the GetQuota operation in version 2022-01-04 to provide more comprehensive quota information. As monitoring shows no external customer calls to the current operation, we are retiring it.

• Impact

  After the retirement date, all calls to these api operations will fail and return the InvalidApi.NotFound or InvalidAction.NotFound error code.

• Recommended actions

  If your code has a dependency on the SDK for these retired api operations, such as specific request or response objects, you may encounter missing dependencies when you upgrade to a newer SDK version. We recommend updating your code to use the replacement api operations or remove the related features. If your code calls these api operations directly, you must also update these calls.

If you have any questions or need assistance, contact us through your dedicated DingTalk group or the MaxCompute developer community (DingTalk group ID: 11782920).

Thank you for your cooperation.

July 4, 2024: Job O&M adds new time parameters

Dear MaxCompute users,

MaxCompute now provides the start time, wait time, and execution duration parameters in the Job O&M list and on the job analysis overview page.

• Start time: The time when a job acquires its first computing resources. For short-running jobs or those that consume no computing resources, such as DDL statements, the job submission time is used instead.

• Wait time: The duration from the job submission time to the start time.

• Execution duration: The duration from the start time to the end time.

These parameters are hidden by default in the Job O&M list. You can click custom columns to display them. These parameters help you quickly determine if a job is running slowly due to long waits for resources. You can then use job analysis to diagnose whether the root cause is insufficient computing resources or resource contention among jobs. Based on this diagnosis, you can optimize job execution, adjust job priorities, or manage computing resources.

For more information, see Job O&M.

If you have any questions or need assistance, contact us through your dedicated DingTalk group or the MaxCompute Developer Community group (DingTalk group ID: 11782920).

Thank you.

July 3, 2024: MaxFrame feature release

Dear MaxCompute users,

The MaxFrame distributed computing framework is now generally available. Effective Wednesday, July 3, 2024 (Beijing time), MaxFrame is available in the China (Shanghai), China (Beijing), China (Ulanqab), China (Shenzhen), and China (Chengdu) regions. Availability will be gradually expanded to other regions.

MaxFrame is a distributed computing framework developed by Alibaba Cloud that provides a Python programming interface and connects directly to MaxCompute's computing resources and data interfaces. This allows Python developers to efficiently process large-scale data and develop AI models. Fully compatible with the Pandas interface, MaxFrame automatically manages distributed processing. This allows you to leverage MaxCompute's massive computing and data resources in a familiar environment for data processing, visual data exploration, scientific computing, and ML/AI development.

For feature details and usage instructions, see MaxFrame: The distributed AI compute engine.

For questions or assistance, contact us through your dedicated DingTalk group, the official MaxFrame user support group (DingTalk group ID: 37130012987), or the MaxCompute developer community group (DingTalk group ID: 11782920).

Thank you.

July 1, 2024: Computing cost optimization in four new regions

Dear MaxCompute users,

The MaxCompute computing cost optimization feature is now available in four new regions: China (Hong Kong), Singapore, Indonesia (Jakarta), and Germany (Frankfurt). This feature analyzes your job workload and target resource configuration to recommend an optimized resource plan for your subscription-based, level-1 quota computing resources. It also projects costs and provides recommendations for switching projects from a pay-as-you-go to a subscription model, helping you optimize computing costs and improve resource utilization. With this release, the feature is now available in a total of ten regions: China (Hangzhou), China (Shanghai), China (Beijing), China (Zhangjiakou), China (Shenzhen), China (Chengdu), China (Hong Kong), Singapore, Indonesia (Jakarta), and Germany (Frankfurt).

For more details, see Computing cost optimization.

For questions, contact us through your dedicated DingTalk group or the MaxCompute Developer Community DingTalk group (ID: 11782920).

Thank you.

May 17, 2024: MaxCompute LogView security upgrade

Dear MaxCompute users,

MaxCompute is rolling out a comprehensive security upgrade for LogView, starting at 14:00 on Monday, June 3, 2024 (UTC+8). This upgrade will be progressively deployed across the following regions: China (Chengdu), China (Zhangjiakou), China (Shenzhen), China (Beijing), China (Shanghai), China (Hangzhou), and China (Ulanqab). After the upgrade, LogView links for jobs submitted from DataWorks DataStudio or DataAnalysis using a shared resource group will redirect to the MaxCompute console. To access a job's LogView, you must meet both of the following conditions:

• You must have a valid Alibaba Cloud account that has completed real-name verification.

• You must be a member of the MaxCompute project where the job runs. A project administrator can add you to the project.

Therefore, if the access identity configured for the MaxCompute engine in a DataWorks workspace differs from your user identity, you may be unable to access the LogView for jobs you submit. This occurs if you are not a member of the MaxCompute project that owns the job. For example, in a standard mode DataWorks workspace, if you submit a job from the development environment to the production environment, you cannot access its LogView if you are not a member of the production environment's MaxCompute project.

To prevent this issue, take the following actions:

• Ensure that all team members who need to access LogView have a real-name verified Alibaba Cloud account.

• Promptly add team members to the appropriate MaxCompute projects and grant them the necessary permissions.

  Note

  For details on user management and authorization operations, see User planning and management and MaxCompute permissions.

During or after the upgrade, if you have questions about LogView access and permission configuration or need technical support, please contact us through your dedicated DingTalk group or the MaxCompute Developer Community group (DingTalk group ID: 11782920). Thank you for your cooperation.

May 13, 2024: MaxCompute etcd certificate upgrade

Dear MaxCompute users,

We will be upgrading the etcd certificate for the cluster in the Singapore region from 10:00 to 16:00 on Monday, May 20, 2024 (UTC+8). This upgrade should not affect your running jobs. However, in the unlikely event that a job fails, please rerun it.

For questions or assistance, contact us through your dedicated DingTalk group or the MaxCompute Developer Community group (DingTalk group ID: 11782920).

Thank you for your cooperation.

May 8, 2024: Job performance observation for MaxCompute

Dear MaxCompute users,

The MaxCompute job performance observation feature will be released at 20:00 on Wednesday, May 8, 2024 (UTC+8). This feature lets you view trends in the number of jobs, CU usage, and job runtimes.

These metrics help you quickly identify issues, such as abnormal spikes in job counts or resource usage, so you can take timely action. By comparing job runtimes over time, you can also pinpoint slow jobs that degrade overall performance.

For more details and usage instructions, see job performance observation.

For questions or assistance, contact us through your dedicated DingTalk group or the MaxCompute developer community group (DingTalk group ID: 11782920).

Thank you.

April 30, 2024: MaxCompute storage upgrade

Dear MaxCompute users,

MaxCompute will undergo a storage upgrade and in the China (Hong Kong) region from 10:00 to 16:00 on Wednesday, May 15, 2024 (UTC+8). This low-risk upgrade should not affect your running jobs. In the unlikely event that a job is affected and fails, please retry it.

If you have any questions or need assistance, you can contact us through your dedicated DingTalk group or the MaxCompute Developer Community group (DingTalk group ID: 11782920).

Thank you for your cooperation!

April 23, 2024: Lifecycle changes for MaxCompute projects in overseas regions

Dear MaxCompute users,

Starting Thursday, April 25, 2024 (UTC+8), when you activate the pay-as-you-go service in China (Hong Kong), Japan (Tokyo), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Germany (Frankfurt), UK (London), US (Silicon Valley), US (Virginia), UAE (Dubai), or SAU (Riyadh - Partner Region), the lifecycle of all projects in that region will align with that of the pay-as-you-go service. For pay-as-you-go usage such as storage and downloads, both the Product Details and the Billable Item (under Usage Records) will be listed as MaxCompute (Pay-As-You-Go). If you purchase subscription resources in that region, only their lifecycle will remain aligned with that of the subscription instance.

For questions or assistance, please contact us through your dedicated DingTalk group or the MaxCompute Developer Community group (DingTalk group ID: 11782920).

Thank you!

April 7, 2024: New MaxCompute subscription discounts

Dear MaxCompute users,

Effective 10:00 (UTC+8) on April 7, 2024, new discounts will apply to various MaxCompute subscription resource specifications, including standard computing resources and the Tunnel service.

The new discounts are: 15% off for a 1-year term; 30% off for a 2-year term; 50% off for a 3-year term.

March 14, 2024: MaxCompute supports ingesting ActionTrail events

Dear MaxCompute users,

You can now create a trail in ActionTrail to deliver ActionTrail events to MaxCompute.

ActionTrail retains events for 90 days. To store and analyze events beyond this period, create a trail in ActionTrail to deliver them to MaxCompute.

For more information, see Deliver ActionTrail events to MaxCompute.

For questions or support, contact us through your dedicated DingTalk group or the MaxCompute Developer Community group (DingTalk group ID: 11782920).

Thank you.

February 29, 2024: PyODPS node upgrade for DataWorks exclusive resource groups

Dear MaxCompute users,

To enhance PyODPS support, DataWorks will gradually upgrade exclusive resource groups in various regions from Monday, April 8, 2024 to Thursday, May 30, 2024 (UTC+8). The following table details the main enhancements and the specific upgrade schedule for each region. The PyODPS package version on the nodes will remain unchanged after the upgrade.

The upgrade may disrupt running PyODPS jobs. If a PyODPS job in an exclusive resource group fails, or if you need help with related issues, contact us through your dedicated DingTalk group or the MaxCompute Developer Community DingTalk group (ID: 11782920).

Thank you for your cooperation!

February 28, 2024: MaxCompute reliability drill

Dear MaxCompute users,

MaxCompute will perform a service access reliability drill in the China (Zhangjiakou) and China (Chengdu) regions from 15:00 to 16:00 on Wednesday, February 28, 2024 (UTC+8). We do not expect this drill to affect your jobs. Should a job fail, please retry it.

For any questions or assistance, please contact us through your dedicated DingTalk group or the MaxCompute Developer Community group (DingTalk group ID: 11782920).

Thank you for your cooperation.

January 25, 2024: MaxCompute computing cost optimization feature released

Dear MaxCompute users,

On Thursday, January 25, 2024 (UTC+8), the MaxCompute computing resource optimization feature has been upgraded to the computing cost optimization feature.

The computing cost optimization feature builds on the existing computing resource optimization feature, which generates optimized resource configuration plans for level-1 subscription quota computing resources. In addition, this upgrade provides a recommended resource configuration plan and a projection of its effects to help you decide whether to switch from pay-as-you-go computing resources to subscription computing resources. This helps you further optimize computing costs and improve resource utilization. For more information and usage instructions, see computing resource configuration optimization.

For questions or assistance, contact us through your dedicated DingTalk group or the MaxCompute Developer Community group (DingTalk group ID: 11782920).

Thank you.