MRS now supports MemArtsStore, T Cloud Public’s high-availability distributed write cache service designed to accelerate big data workloads.

• Provides remote shuffle services for big data applications.
• Eliminates dependence on local disks for shuffle data storage.
• Delivers high availability through multiple data replicas.
• Prevents shuffle data loss in the event of a node failure.
• Optimizes performance by aggregating data within partitions and reducing small I/O operations.

This enhancement improves the resilience and efficiency of large-scale data processing workloads.

Apache Doris now supports the creation and execution of custom user-defined functions.

• UDF (User-Defined Function)
• UDAF (User-Defined Aggregate Function)
• UDTF (User-Defined Table Function)

Using Java-based APIs, customers can implement custom business logic directly within Doris, enabling greater flexibility for analytics and data processing scenarios.

Hive now supports permission-based visibility for databases and tables when Ranger authentication is enabled.

• Users can view only the databases and tables they are authorized to access.
• Unauthorized databases and tables remain hidden.
• Improves security and user experience by reducing exposure to irrelevant metadata.

This feature helps organizations implement stricter access controls and governance policies.

Hive now allows administrators to select the Java runtime used for job execution.

• JDK 8 (Standard)
• JDK 17
• JDK 21

This capability is available for Hive jobs running on both MapReduce and Tez execution engines, providing greater flexibility for application compatibility and modernization initiatives.

Hive clients now support username and password authentication when connecting to HiveServer.

• Simplifies client setup and connectivity.
• Eliminates the need to configure Kerberos client files such as krb5.conf, keytab, and hosts.
• Remains compatible with clusters operating in Kerberos-enabled security mode.

This enhancement reduces operational complexity while maintaining secure access to Hive services.

Dynamic resource plans have been enhanced with scheduled policy support.

• Defining different resource allocation strategies for different time periods.
• Automatically adjusting resource plans based on business schedules and workload patterns.
• Improving resource utilization and cluster efficiency throughout the day.

This feature enables more flexible and intelligent resource management.

HBase now supports PhoenixQueryServer instances.

• Enables SQL-style access to HBase data through Apache Phoenix.
• Supports connectivity using lightweight Phoenix clients.
• Simplifies analytical and application integration scenarios.

This enhancement makes HBase data more accessible for developers and analysts.

FlinkServer now supports scheduled job optimization, allowing automatic adjustment of operator parallelism at predefined times.

• Dynamically updates operator parallelism to match workload demands.
• Supports SQL streaming jobs running in active/standby deployments.
• Up to five optimization plans can be configured per job.
• Enables improved resource utilization and workload efficiency.

• Supported only for jobs in the Running state.
• Not available for active-active deployments.
• Available exclusively for jobs submitted through FlinkServer.
• Uses the adaptive scheduler when scheduled optimization is enabled.
• The maximum supported operator parallelism is 32,768.

This feature helps customers optimize streaming workloads while reducing manual operational effort.

Further information can be found in the MapReduce Service (MRS) area in the Help Centers.