Feature Updates (2025)

更新时间: 2026-03-20 07:18:43

This topic provides updates on Dataphin shared mode (fully managed edition) releases in 2025.

December 2025

Product version: V5.3.4

Beijing and Chengdu: December 2, 2025.

Shenzhen and Hangzhou: December 4, 2025.

Shanghai: December 7, 2025.

Feature name

Description

Related documentation

Data source management

OpenSearch data sources are now supported.

Create OpenSearch data sources

Data Service Studio

Asynchronous calls to MaxCompute direct connection data source APIs now return full data.

Create APIs using direct connection data sources (query type)

November 2025

Product version: V5.3

Beijing and Chengdu: November 4, 2025.

Shenzhen and Hangzhou: November 6, 2025.

Shanghai: November 9, 2025.

Feature name

Description

Related documentation

Approval templates

Approval templates now support asset listing and delisting scenarios.

Create and manage approval templates

Data source management

  • DolphinDB data sources are now supported.

  • Microsoft SQL Server data sources support specifying InstanceName. When InstanceName is configured, the JDBC URL connects without a port number.

  • Redis data sources now support version 5.x.

Permission management

The system alerts you if your account lacks row-level permissions for any objects when you run or submit integration or computing tasks. Contact your security administrator to obtain authorization.

Projects

Code review is now supported for Basic and Dev-Prod projects. After enabling code review, offline computing tasks in the project require code review before submission.

Create general-purpose projects

Attribute management

Tables, business metrics, technical metrics, APIs, and dashboards now support two custom attribute entry methods: dropdown multi-select and hyperlinks. When using dropdown single-select or multi-select, enumeration values can reference system attributes.

Create and manage attributes for assets, APIs, and dashboards

Offline integration

  • Offline integration input components and whole-database migration source endpoints now support DolphinDB data sources.

  • Redis output components support batch data writes.

  • Kafka output components support one-click Topic creation.

  • Spark writes to Hudi now support batch writes. Data is written to the target table in batches according to the specified ratio.

Offline development

  • You can select download formats when downloading sample data.

  • Database SQL tasks now support OceanBase data sources and full data downloads.

  • MaxCompute computing tasks support setting default lifecycles for tables.

Real-time development

  • When selecting Kafka as the data source for metadata tables, protobuf message body format is supported.

  • Flink SQL views now support the lateral join syntax.

Create and edit offline physical tables

Data Quality

TDSQL for MySQL data sources are now supported for quality monitoring.

Metadata Center

DolphinDB, openGauss, and TDSQL for MySQL are now supported as metadata acquisition sources.

Create and manage metadata acquisition tasks

Data Service Studio

  • You can register Large Language Model (LLM) APIs. Register LLM services in Dataphin Data Service Studio for centralized management. Track model usage across applications, apply rate limiting, monitor calls, and view logs.

  • Default values, descriptions, and example values for API input or output parameters now support up to 1000 characters. Input parameter length has no limit.

  • API registration now includes testing. Verify that responses match expectations. When configuring response parameter paths, test responses auto-suggest available data.

  • Calling APIs via SDK or debugging or testing them in Dataphin Data Service Studio automatically triggers the GetJobExecutionLog API. If the call fails, you can view detailed error logs.

  • DataService Studio Application Optimization:

    • Application users must obtain approval before creating applications.

    • Data Service Studio applications are not tied to specific Data Service Studio projects. Users can request API permissions from all projects.

  • System administrators can configure network settings, system settings, project management, and application management.

  • Direct connection data source APIs now support OceanBase and Databricks data sources. Service units now support OceanBase data sources.

Catalog management

  • Approval rule functionality is added. Before listing assets, rules check whether attributes are complete. Before delisting, rules assess business impact. This creates a structured, controllable environment for asset management.

  • Business metrics, APIs, and dashboards now support batch import and export.

  • Display names and descriptions for table and field assets update automatically when DDL comments change.

September 2025

Product version: V5.2

Beijing and Chengdu: September 18, 2025.

Shanghai: September 21, 2025.

Shenzhen and Hangzhou: September 23, 2025.

Feature name

Description

Related documentation

Super X

Super X intelligent applications launch with core features: X-Data Engineering, X-O&M Assistant, X-Coding Assistant, and X-Analytics. These provide end-to-end intelligent solutions for data development, governance, and operations. They reduce manual effort, improve business responsiveness, and increase data processing accuracy.

Super X (intelligent application)

Basic settings

In Management Hub > System Settings, basic settings are added. Configure tenant-level scheduling time zones. Tasks then schedule according to this time zone.

Basic settings

Data source management

  • TDSQL for MySQL data sources are now supported.

  • Oracle data sources in host mode support multiple IPs.

  • TDengine data sources now support version 2.2.1.3.

  • API data sources support custom token prefixes in Authorization headers.

  • Default time zones can be set for MySQL, Oracle, Microsoft SQL Server, PostgreSQL, AnalyticDB for MySQL 2.0, AnalyticDB for MySQL 3.0, AnalyticDB for PostgreSQL, Amazon Redshift, and Databricks data sources. Time format data in integration tasks is processed using these time zones.

Permission management

Offline computing function permissions can now be controlled at the project level. After enabling, cross-project references to custom functions require explicit permission requests.

Offline development

  • Offline scheduling templates are now supported. Computing tasks, integration tasks, logical tables, derived metrics, and offline datasets follow referenced scheduling templates.

  • Database SQL tasks now support DM (Dameng) and openGauss data sources.

  • MaxCompute now supports SELECT * except(<column>) and SELECT * replace(<column>) syntax.

  • Full data downloads and temporary tables created by Notebook tasks use the default storage format set in Management Hub > System Settings > Development Platform.

Offline integration

  • TDSQL for MySQL data sources are now supported for input/output components and whole-database migration sources.

  • Databricks output components support Overwrite and Update load strategies. Microsoft SQL Server, AWS RDS for SQL Server/PostgreSQL, PostgreSQL, and Dameng (DM) output components support Update on primary key conflict.

  • For API input components with multiple requests, choose between Pagination loop and Parameter iteration loop modes.

  • Time-based data is converted across time zones using time zone information from input and output component data sources.

Metadata Center

SelectDB, Doris, and DM (Dameng) are now supported as metadata acquisition sources.

Create and manage metadata acquisition tasks

Asset catalog / Catalog management

Business metrics now support generating layered, drill-down metric relationship diagrams using logical operation expressions based on related business metrics.

Data Service Studio

  • Gateway timeout configuration is added to monitor API call duration.

  • Data Service Studio now supports Amazon Redshift and Databricks data sources. Amazon Redshift supports synchronous and asynchronous calls for direct connection data source APIs and service units. Databricks supports only asynchronous calls for direct connection data source APIs.

  • Direct connection data source APIs and service unit APIs support row-level permission controls. Row-level permissions inherit table-level permissions. Two authentication modes are supported: application proxy and personal account.

June 2025

Product version: V5.1

Hangzhou: Released on June 24, 2025.

Shenzhen: Released on June 26, 2025.

Beijing: Released on June 29, 2025.

Shanghai: Released on July 1, 2025.

Feature name

Description

Related documentation

Member management

A new permission point, Enable cross-tenant publishing mode, is added. Users with this permission can enable cross-tenant publishing in O&M mode.

Appendix: Global and project role permissions

Permission management

You can now combine multiple fields in the same table for granular control of sensitive data security.

Add and manage row-level permissions

Data sources

  • Data sources can be saved without passing connection tests. The system runs tests in the background.

  • API data sources support OAuth 2.0 authentication and dynamic token acquisition during API calls.

  • Default time zones can be set for MySQL, Oracle, Microsoft SQL Server, PostgreSQL, AnalyticDB for MySQL 2.0, AnalyticDB for MySQL 3.0, AnalyticDB for PostgreSQL, Databricks data sources, and compute sources. Integration tasks process time-format data using these time zones. This handles international deployments where time zones differ between regions, data sources, and Dataphin servers.

  • Amazon RDS for PostgreSQL/MySQL/SQL Server/Oracle/DB2 and Amazon Redshift data sources are now supported. Use them for offline integration, metadata acquisition, cross-domain table quality rules/data source monitoring rules, and API development. (Redshift and DB2 do not yet support Data Service Studio.)

Cross-tenant publishing

When publishing data sources across tenants, you can now set JDBC URL replacement rules in batches.

Manage global pending publish objects

Project management

When configuring release approvals for Dev-Prod projects, you can reference custom approval templates to flexibly define approval nodes.

Create general-purpose projects

Compute Source

Ververica Flink compute sources support manually entering region IDs. This standardizes region entry across different Dataphin deployment modes.

Create Ververica Flink compute sources

Offline integration

  • Whole-database migration supports setting scheduling resource groups for development tasks.

  • Offline pipeline input/output components and whole-database migration source endpoints now support Amazon Redshift, Amazon RDS for PostgreSQL, Amazon RDS for MySQL, Amazon RDS for SQL Server, Amazon RDS for Oracle, and Amazon RDS for DB2 data sources.

  • When Hive input/output components read or write Iceberg-formatted tables, partitioned tables are hidden. When Hive output components select Hudi lake tables, one-click Hudi table creation is supported.

  • One-click table creation for AnalyticDB for PostgreSQL output components supports the DISTRIBUTED BY statement.

  • API input/output components support custom signature rules. Generate dynamic signatures using MD5HEX and HMAC functions with timestamps, random numbers, and request parameters.

  • Offline integration supports data processing using time zones configured for data sources or compute sources.

Offline development

  • Running tasks in the development environment requires global variable permissions.

  • In Dev-Prod environments, objects published to production (computing tasks, integration tasks, logical tables, metrics, etc.) must first be unpublished in development. Then, unpublish the object in the pending publish list to remove it from production. Alternatively, keep the original configuration in the development platform.

  • Configure automatic retries for task timeouts.

  • New computing tasks support default top-of-file comments.

  • Database SQL tasks now support Microsoft SQL Server and Hologres data sources.

  • GaussDB (DWS) database SQL tasks continue execution after syntax errors. Original SQL runs using data source execution permissions.

  • Database SQL tasks can be created for data sources unreachable from Dataphin's default cluster.

Real-time development

When selecting Hudi as the data source for metadata tables, primary keys and partition fields are auto-detected. You no longer need to specify them manually in Flink SQL.

Create and manage metadata tables

Task O&M

  • O&M Center removes project isolation. View all authorized tasks and instances at once. Filter by project.

  • Recurring instance lists now show instance ID, retry count, and auto-retry count. Recurring task details show code versions and version comparisons. Offline instance details show CPU and memory consumption.

  • AI for IT operations must be enabled. For recurring instances, view Gantt charts for any two upstream/downstream instances in the DAG diagram.

  • You can modify instance priorities for recurring and backfill instances.

  • Download recurring instance lists or filtered lists.

  • Run manual tasks for date ranges to generate multiple manual instances at once.

  • A tenant-level scheduling resource dashboard is added. View Dataphin cluster resource allocation and consumption trends. Identify and export under-optimized tasks with abnormal resource consumption/assignment ratios for tuning.

  • Real-time task alert frequency is reduced to once per minute to avoid spam.

Asset inventory

After enabling data source table exploration in Administration > Metadata > Explore and analyze, start exploration and view reports from the asset inventory details page.

Data source table details

Data standards

The Import from table method is added for creating lookup tables. Reference compute sources or collected data source tables with SELECT permissions to quickly create lookup tables. Edit retrieved codes manually or refresh all codes using the referenced configuration.

Create and manage lookup tables (code tables)

Data Quality

  • When adding cross-database/schema monitoring objects for the same data source, configure them in batches to improve efficiency.

  • All data source types support Data source connectivity monitoring rules.

  • Data Quality supports monitoring rules for data source tables unreachable from Dataphin application clusters. (Use with registered scheduling clusters and metadata acquisition.) SAP HANA and IBM DB2 are not supported.

Metadata acquisition

Data source table exploration configuration is added. Configure exploration by data source granularity, including concurrency throttling, exploration enablement, and timeout. Set SQL statements to tune exploration tasks.

Asset catalog

  • View business metric lists. Business and technical metrics both belong to the Metric asset type. Filter by metric type.

  • Technical metrics now display data type, measurement unit, cumulativeness, and related business metric attributes.

  • View exploration reports on data source table and custom metric asset detail pages, based on listing configurations.

  • Standard modeling metric detail pages include metric relationship diagrams. These visualize metric decomposition clearly for business users.

Catalog management

  • Create business metrics and manage their listing.

  • Technical metrics now support data type, measurement unit, cumulativeness, and related business metric attributes.

  • Data source tables and custom metrics support exploration report permission controls.

Manage quasi-assets

Data Service Studio

  • MaxCompute data sources are now supported for asynchronous API creation using direct connection data sources.

  • API testing/debugging now supports asynchronous APIs. Logs appear in call detail logs and can be monitored in the O&M dashboard.

  • API names can now be edited. Changes take effect globally and immediately. API documentation Hosts now include domain names. When creating APIs using direct connection data sources or SQL mode, request and response parameters support batch operations. SQL trial run results can populate parameter example values.

April 2025

Product version: V5.0

Beijing: Released on April 15, 2025.

Hangzhou and Shenzhen: Released on April 17, 2025.

Shanghai: Released on April 20, 2025.

Feature name

Description

Related documentation

Sales

RAM users with AliyunDataphinFullAccess and AliyunBSSOrderAccess permissions can activate, upgrade, and renew Dataphin instances.

AI assistant

An AI assistant help entry is added. Quickly launch the AI assistant.

Support

Smart generation

Catalog management adds Smart Generation (public preview). Using Large Language Models, it intelligently generates descriptions, tags, and project assignments. It auto-detects custom metrics in tables and creates metric basics with one click. (Free during public preview. Billing starts after commercialization.)

Manage quasi-assets

Global role management

Built-in system roles (except super administrator) support editing permissions.

Global role management

Resource settings

Set upper and lower limits for available resources in each scheduling resource group. Idle resources in a group can be borrowed by other groups when needed.

Create custom resource groups

Data source management

  • Custom data sources (non-relational databases) support JAR packages up to 200 MB.

  • When creating Doris data sources, HTTPS protocol is supported for FE nodes.

  • Set default resource groups for data sources. These groups are used by default for database SQL, offline whole-database migration, and data preview tasks.

Data source management

Common business logic

Any date in the common calendar supports updating its date tag.

None

Offline integration

  • TDengine and Databricks data sources are now supported for offline integration.

  • Whole-database migration supports TDengine as a source database and Databricks as a target database.

  • Offline whole-database migration supports external data sources as source data. (First, configure metadata acquisition tasks for the data source.)

Offline development

  • When the compute engine is MaxCompute and the storage class is internal table, Delta and transactional tables are editable.

  • Database SQL tasks now support Doris and SelectDB data sources.

  • Objects in the left directory of Data Development now support up to 10,000 items.

  • MaxCompute now supports json literal constants, such as json '{"key":"value"}'.

  • Offline task scheduling dependencies now support Nearest previous instance and Nearest next instance.

Basic O&M

  • Backfill tasks now support custom business dates. Specify a single day or quickly select month-end dates.

  • For backfill tasks with hourly or minute-level schedules, specify a time range to generate instances only for that period.

Create backfill tasks

Asset inventory

Data sources supporting metadata acquisition show field lineage generated by integration tasks. All data sources support table lineage.

Data standards

  • Standard implementation monitoring configuration is optimized. For system attributes, choose whether to add built-in implementation monitoring.

  • Edit data standards and simultaneously edit linked standards and documents.

Data security

  • Add classification and grading identification results for data source tables. Add manually or in batches.

  • Rule scanning configuration now supports choosing whether to scan views. If disabled, you can still add view identification results manually.

  • Manually clear identification results. When deactivating a classification, choose whether to delete existing identification results and records.

  • Run individual identification rules manually.

Metadata acquisition

  • ClickHouse, Greenplum, and OceanBase data sources are now supported as metadata acquisition sources.

  • Quick BI dashboard acquisition is now supported. View dashboard basics and properties. List dashboards in the asset catalog.

Catalog management

  • Asset topics support batch editing names and descriptions.

  • Quick BI dashboard listing is now supported.

  • When manually listing table assets, batch-edit related metrics' catalog, tags, and statistical periods. Add all required attributes with one click.

  • Preview data source table data in the asset catalog.

  • When editing assets, usage instructions support adding images and videos.

Data Service Studio

  • GaussDB (DWS) data sources are now supported for direct connection data source APIs.

  • Service unit lists now show API details for associated service units.

  • Customize API data update frequency.

Tag Factory

  • Offline datasets and behavioral relationships support logical tables as data sources.

  • Date-type tag condition configuration is enhanced:

    • For rule-combined tags and audience groups, dynamic time conditions now support Before, On or before, After, and On or after. These work with past and future times.

    • For behavioral preference tags, behavioral statistics tags, rule-combined tags, and audience groups, dynamic time ranges now support shortcuts like This year, This month, and This week. Dynamic time is calculated based on bizdate.

    • The tag and audience group directory tree in the asset marketplace now shows all nodes. Total asset counts reflect filtered results.

    • When creating offline tag services, offline audience group services, and paginated audience group services, the owner dropdown supports search.

    • Real-time tag version lists now include a Test button. Testing is required before submission.

OpenAPI

Data Service Studio OpenAPI is now supported.

OpenAPI portal - Dataphin

March 2025

Product version: V4.5.3

Beijing, Hangzhou, Shenzhen, and Shanghai: Released on March 20, 2025.

Feature name

Description

Related documentation

Management Console

Dataphin shared mode (fully managed edition) adds two RAM management permissions:

  • DataphinOpenAPIReadOnlyAccess: Call read-only Dataphin OpenAPIs.

  • DataphinOpenAPIFullAccess: Call all Dataphin OpenAPIs and modify console AccessKey and AccessSecret.

Note

In addition to having one of these permissions, you need the corresponding API operation permissions to call Dataphin OpenAPIs.

Management Console overview

Metadata Center

Metadata change version tracking is added. Save historical versions and version counts for compute source tables and data source tables to reduce storage costs.

Metadata change records

February 2025

Product version: V4.5

Beijing: Released on February 11, 2025.

Hangzhou and Shenzhen: Released on February 13, 2025.

Shanghai: Released on February 16, 2025.

Feature name

Description

Related documentation

Management Console

Dataphin Management Console AccessKey and SecretKey configuration requires ListUsers and GetAccountAlias permissions.

Management Console overview

Project role management

Custom project roles include Project Quality Management – Quality Rule Management permissions. System roles (except tag project roles) support editing, adding, and removing Project Quality Management – Quality Rule Management permissions.

Appendix: Global and project role permissions

Global role management

Custom global roles include Quality Rule – Manage, Rule Template – Manage, and Knowledge Base – Manage permissions.

Data sources

You can now test Greenplum and FTP data source connectivity with external scheduling clusters.

Approval templates

An approval template for owner-only key management is added for registered keys where only the owner can manage them.

View and configure built-in approval templates

Offline integration

  • Whole-database migration targets support reading and writing MaxCompute Delta tables. SelectDB and Doris data sources are added, with one-click table creation.

  • For MySQL input components with multiple source tables, you can select multiple databases and tables to sync data. This supports sharding scenarios.

  • Logical table input components support selecting and deselecting all foreign key fields.

Offline development

  • Database SQL tasks now support AnalyticDB for MySQL 2.0, AnalyticDB for MySQL 3.0, and GaussDB (DWS) data sources.

  • For hourly scheduling cycles, you can choose from time ranges, exact hours, custom time points, or custom time ranges.

  • Conditional scheduling is upgraded. Control instance scheduling for different dates and times using runtime date and time. Preview scheduling plans by runtime date.

  • After submitting PyODPS tasks to MaxCompute, the system automatically reports metadata matching SQL tasks. You can locate tasks in the MaxCompute console for cost tracking.

  • When running manual tasks in production, global variables automatically substitute values.

  • When running code in development, global variable values for accounts and passwords are masked in logs.

  • Online Python package installation or upgrades support specifying module versions.

Asset inventory

The lineage diagram is upgraded. You can view lineage for objects with circular dependencies and quickly search nodes in expanded lineage diagrams.

Dataphin asset details

Data security

  • When inheriting identification results from lineage, only direct lineage is used. Indirect lineage is filtered to improve accuracy.

  • Character limits for data classification definitions are increased: classification names support 128 characters, abbreviations support 64 characters, and descriptions support 2048 characters.

Metadata Center

  • MySQL, AnalyticDB for MySQL 3.0, PolarDB-X (formerly DRDS), and StarRocks data sources now support collecting all databases or specific databases in a single source.

  • Source systems support configuring system types for building system lineage.

Asset catalog

  • An API object list is added. You can view APIs by topic or catalog.

  • Asset filtering is optimized. Combine object-type attributes with filters for flexible, fast asset searches.

  • Table and metric asset lists are optimized. Customize displayed columns to get more information quickly and simplify asset lookup.

  • When editing table and metric assets, you can enable lineage, quality overview, and metadata changes (tables only). View related information in the asset catalog details:

    • Lineage: View system, table, and field lineage diagrams. Switch to view tables in a system node or fields in a table node. Hide nodes that are not yet listed or are listed but inaccessible. View summaries for listed tables and metrics. Jump to details quickly.

    • Quality overview: View quality scores for tables and metrics across six dimensions. See rule statistics and validation records.

    • Metadata changes: View change records between any two versions of a table. If the latest change includes schema evolution, see potential downstream impacts to notify stakeholders.

Catalog planning

Batch import and export for asset topics is added. Use system templates. Fill in configurations to import offline content into Dataphin. Or download existing catalog configurations, edit them locally, then re-import.

Batch import/export asset topics

Asset listing – manual listing

Batch import and export for table and metric assets is added. Download Excel templates for current configurations. Update them, then re-import into Dataphin. For failed imports, download error logs with one click from batch operation records.

Batch import/export assets

Asset listing – automatic listing

  • A new rule type, Do not list, is added. Automatically move assets from quasi-asset to do-not-list status for specified scopes. This reduces asset management complexity.

  • You can manually run or stop individual automatic listing rules.

  • Guidance on configuring and enabling automatic listing rules is added.

Data Service Studio

  • Script mode APIs are optimized. Support single-value and multi-value parameters. For LIKE operators, support fuzzy match, right match, and left match.

  • APIs support joint debugging and testing before publishing.

  • Call instructions are optimized. Show document outlines. Upload or hide Java SDK examples and call instructions.

  • For direct connection data source and SQL mode APIs, trial run results auto-populate API example values.

  • You can switch between APIs quickly on API view, edit, test, debug, and documentation pages.

  • You can batch publish APIs. Logical table APIs are excluded.

Tag Factory

  • Offline tags now support computed tags. Build tag logic using relationships and functions on existing tags. Visualize tag computation results as tag values.

  • Rule-combined offline tags support nested combinations for secondary processing.

  • Offline tag services support exporting behavioral preference tags and their linked lookup table information.

  • Offline tag and audience group services support batch export of selected tag values or code descriptions.

  • Offline dataset scheduling dependencies support adjusting dependency policies, such as All instances or Last instance.

  • Lookup table code values now support up to 500 entries.

Cross-tenant publishing

When publishing Python-based tasks across tenants, use the same Python version and module name as the target tenant. Skip publishing Python packages across tenants.

None

上一篇: Feature Updates (2026) 下一篇: Feature Updates (2024)
阿里云首页 智能数据建设与治理 Dataphin 相关技术圈