Acceldata ODP 3.3.6.4-1 Release Notes

This release significantly expands ODP's real-time analytics, machine learning, and streaming capabilities, introducing six major new integrations, broader stack coverage with the latest Apache versions, and deeper connector support across Spark 4 and Kafka. Alongside these capability additions, this release delivers substantial security hardening with 550 CVE fixes across the stack. Customers can now build modern, decoupled data and AI workflows on a unified, Ambari-managed platform.

Key Highlights

• New Platform Integrations: ODP now supports the following integrations to enhance real-time analytics and machine learning workflows:

  • Apache Celeborn 0.6.2 — Remote Shuffle Service: ODP now supports Apache Celeborn, a high-performance Remote Shuffle Service (RSS) for distributed compute engines. Includes full lifecycle support: deployment, configuration, security, monitoring, performance tuning, and storage management with native integration for Spark, Flink, Tez, and MapReduce.
  • Spark Connect — Decoupled Client-Server Architecture: ODP now supports Spark Connect, enabling remote Spark application development with improved stability, scalability, and upgrade flexibility. Supports interactive debugging and both PySpark and Scala applications, freeing client environments from tight server coupling. Available across Spark 3.5.1, 3.5.5, and 4.1.1.
  • Apache Superset 6.0.0 — Modern BI & Visualization: Added support for Apache Superset deployment in Ambari-managed ODP environments, with end-to-end support for installation, configuration, UI access, connector management, administration, and clean uninstallation. For details, see Install Superset.
  • Apache Spark 4.1.1: ODP now supports Apache Spark 4.1.1 with major enhancements across Streaming, SQL, PySpark, Connectivity, Data Sources, Custom Functions, and Usability. For details, see Apache Spark 4.1.1.
  • Apache NiFi 2.7.2 — Next-Generation Dataflow: ODP now supports the NiFi 2.x line, featuring a modernized UI, Python-based processors, parameter providers, and significant performance and security improvements over the 1.x line. For details, see 3-Node NiFi 2.7.2 Standalone Cluster Setup Guide.
  • Apache Ozone 2.1.0 — Scalable Object Storage: ODP now supports the Ozone 2.x line, delivering improved scalability for billions of objects, enhanced S3 gateway compatibility, stronger multi-tenancy and security controls, and operational improvements over Ozone 1.x. For details, see Install Ozone 2.

• Stack Version Expansions: Customers can select the component versions that best fit their environment. The following versions are newly available in this release alongside previously supported versions:

• Spark 4 Connector Ecosystem: Spark 4.1.1 ships with first-class connectivity to the broader ODP stack:

  • Spark 4 Hive Warehouse Connector — unified access to Hive-managed tables.
  • Spark 4 HBase Connector — direct read/write integration with HBase.

• Platform Capability Extensions: The following new platform-level capabilities are added in this release:

  • HBase Thrift Server — now supported as a managed component within the HBase service, enabling cross-language client access (Python, Ruby, C++, PHP, etc.) to HBase.
  • Oozie–Spark 3 Integration Extension — orchestrate Spark 3 jobs natively through Oozie workflows.
  • Multi-NameNode (HA) Support — high-availability support for clusters running multiple NameNodes, improving resilience and federated namespace operations.

• High Availability (HA) Expansions: HA is now supported for the following services, eliminating single points of failure for critical metadata and orchestration layers:

  • Trino — coordinator HA for resilient query execution.
  • Schema Registry — HA for Kafka schema management.
  • NiFi Registry — HA for versioned dataflow configuration.

• Kafka Tiered Storage with S3 Support: Added support and configuration guidance for enabling Kafka Tiered Storage with S3, allowing Kafka log segments to be offloaded to S3-compatible object storage — reducing broker storage costs while preserving long-retention use cases. Includes Ambari setup, secure credential management, topic-level configuration, and validation of data offloading to object storage. The Kafka AWS Sink Connector is also now available, enabling streaming data delivery directly into AWS services. For details, see Configure Kafka Tiered-Storage with S3.

• Enhanced JupyterHub — Unified Analytics & AI Workspace: JupyterHub is now a complete data science environment with:

  • Multi-kernel support — PySpark, Scala, R, and SQL, switchable at the notebook level.
  • Multi-Spark version support — work across Spark 3.3.3, 3.5.1, 3.5.5, and 4.1.1 within a single workspace.
  • Pre-integrated ML & application libraries — Ray, TensorFlow, PyTorch, and Streamlit ready out of the box, enabling scalable data engineering, ML experimentation, and interactive application development under a single managed environment.

• CVE Fixes & Security Hardening: This release addresses 550 CVEs across the ODP stack, with the highest concentration of fixes targeting Hive, Spark 3, Oozie, and ClickHouse.

For details, see Fixed CVEs.

New Features

• Discover the innovative features in the ODP 3.3.6.4-1 release with upgraded Apache components and platform modernization for enhanced performance, efficiency, and compatibility. For details, see New Features.

Enhancements

• Platform Improvements: This release introduces new features and optimizations that boost the platform's functionality and user experience, enhancing your data management capabilities. For details, see Enhancements.

Resolved Issues

• Bug Fixes: Acceldata's commitment to a seamless data observability experience includes promptly addressing reported issues. This release resolves multiple bugs, ensuring a more stable and reliable platform. For details, see Bug Fixes.
• CVE Fixes: 550 CVEs have been addressed in this release, including 71 Critical and 327 High severity issues, to significantly improve platform security posture. For details, see Fixed CVEs.

Business Impact

• Stronger security posture — 550 CVE fixes including 71 Critical and 327 High severity, reducing risk exposure across the data platform.
• Higher availability — HA for Trino, Schema Registry, NiFi Registry, and Multi-NameNode strengthens production resilience.
• Lower TCO — Kafka Tiered Storage and Celeborn shuffle service reduce infrastructure costs.
• Modern ML/AI readiness — JupyterHub with Spark 4, Ray, TensorFlow, and PyTorch on a governed, unified platform.
• Deployment flexibility — customers can select the component versions that best fit their environment.
• Modern BI — Superset 6.0.0 brings open-source visualization natively into the ODP stack.