Announcing Confluent Platform 8.3: Powerful Apache Flink® SQL operations, Easier KRaft Migrations, Expanded Monitoring and more.

Platform teams want to maximize the value of data in motion, but separate workstreams for stream processing, monitoring, KRaft migrations, and data governance create friction as environments grow. As a result, teams spend more time managing operational overhead and less time building.

Today, we’re excited to announce Confluent Platform (CP) 8.3.0, built on Apache Kafka® 4.3.0, reinforcing our core capabilities as a data streaming platform. This release helps platform teams simplify Apache Flink® SQL operations, migrate their streaming infrastructure with ease, maintain unified visibility while reinforcing security and governance. Explore what’s new in Confluent Platform 8.3:

• Powerful & Simplified Flink SQL Operations with Confluent Platform for Apache Flink 2.4

• Expanded Monitoring via Unified Stream Manager (USM)

• Migrate to KRaft with Greater Ease and Confidence

• Structured, Centralized-governed Data in Motion

You can find additional details about the features in the release notes. For more details about Apache Kafka 4.3.0, read the release blog post.

Powerful & Simplified Flink SQL Operations

Confluent Platform 8.3 is released along with Confluent Platform for Apache Flink version 2.4.0, which continues to make your Apache Flink® operations easier to manage and run with MCP and built-in UI, yet more powerful to use in production by embedding more complex custom logic and reusable artifacts in Flink SQL. 

MCP server for Confluent Platform for Flink

As teams adopt AI agents like Claude Code or Codex for operational tasks, they need those tools to work through their existing control plane. Without a native interface, teams are left with ad hoc ways to connect agents to Confluent Platform for Apache Flink with no governed ways to inspect Flink environments, reason over application state, or safely take action.

The MCP server for Confluent Platform for Apache Flink solves this by bringing an agent-ready interface directly to Confluent Manager for Apache Flink 2.4.0 (CMF 2.4.0), with over 16 read-only debugging tools and 20 write tools for managing resources. Using simple prompts in natural language, AI agents can inspect environments, applications, statements, compute pools, logs and savepoints directly. AI agents run within your existing CMF security guardrails, including current authentication methods and RBAC enforcement. 

UDFs, Artifact Management, and Custom Catalogs for SQL

Today, you can easily write SQL statements with Flink SQL using Data Definition Language (DDLs) and changelogs, and with Kafka topics in CP materialized as tables that can be queried immediately. But as data engineers look to bring custom business logic, external sources and sinks, and more specialized use cases into their streaming applications, they want to do so without stepping outside the SQL workflow.

With User-defined Functions (UDFs), Artifact Management, and Custom Catalogs, you can build richer Flink SQL applications without leaving the SQL workflow.

• UDFs: Bring custom logic into Flink SQL with Java-based functions that can be registered and used directly in SQL workflows.

• Artifact Management: Store, version, discover, and reuse SQL extensions across environments.

• Custom Catalogs: Extend Flink SQL to work with sources and sinks (such as Amazon S3 or blob storage) beyond Kafka, across a broader set of environments.

Embedded Meta Data Service (MDS) in Confluent Manager for Apache Flink

Teams want to get more value from Confluent Platform for Apache Flink through its tight integration with Kafka, including capabilities such as materializing topics as tables in SQL and applying consistent RBAC and authentication across both environments. As teams expand how they deploy and manage Flink, they need a more flexible way to secure Confluent Platform for Apache Flink that aligns with their operating model, reduces overhead, and supports scale across diverse deployment environments. 

Embedded MDS in CMF allows teams to secure and manage RBAC and authentication directly from the Confluent Platform for Apache Flink control plane, giving them an access control model that fits more naturally across different deployment environments. You can now integrate Confluent Platform for Apache Flink with any version of Confluent Platform (7.6 or later), Confluent Cloud, Warpstream or any flavor of Kafka implementation when using RBAC and authentication. This provides greater operational independence, lower infrastructure overhead, and increased flexibility as you scale. 

Confluent Manager for Apache Flink (CMF) UI

As teams expand how they deploy and manage Confluent Platform for Apache Flink, they want an operational experience that is purpose-built for Flink. The CMF UI gives teams a dedicated, single-pane-of-glass UI for Flink operations. Starting with Confluent Platform for Apache Flink 2.4.0, the CMF UI replaces the earlier Flink UI previously provided by Confluent Control Center(C3), as the integration is now deprecated. The CMF UI provides a streamlined and scalable way to manage thousands of resources across environments, applications, SQL statements, savepoints, artifacts, and more. With support for multi-cluster Kubernetes management, artifact management, savepoint management, and native SQL authoring, the CMF UI delivers a purpose-built experience designed to operate large-scale Flink deployments with complete operational visibility.

Periodic Savepoint

Savepoints are a critical part of how teams protect stateful Flink applications and maintain recovery readiness for planned operations, disaster recovery, and point-in-time recovery. Confluent Platform for Apache Flink already gives customers managed savepoint capabilities, but as Flink deployments scale, teams increasingly want a simpler way to operationalize savepoints.

Periodic Savepoints in Confluent Platform for Apache Flink automates savepoint scheduling, retention, and cleanup to simplify Flink recovery operations at scale. Teams can schedule savepoints for Flink applications and statements to maintain recovery points for disaster recovery and point-in-time recovery. You can apply recurring schedules and environment defaults to manage savepoints consistently at scale. Additionally, you can control retention by using policies, restricted windows, and bulk cleanup with more flexibility and less manual effort.

Expanded Monitoring with Unified Stream Manager (USM)

USM unifies how you manage and monitor all your Confluent clusters across on-prem, private cloud, or public cloud deployments within a single unified view. Confluent Platform 8.3 includes USM Agent version 1.2.0, which adds RHEL 8 support, client-side observability, and centralized visibility to troubleshoot Kafka Streams applications faster.

Kafka Streams (KStreams) UI on USM

KStreams is widely used for critical stream processing workloads. Operators have needed a unified way to understand application health and performance across Confluent Platform environments, spot issues quickly and troubleshoot with confidence. 

KStreams UI is now available on USM, bringing intuitive application visibility and troubleshooting for KStream applications to Confluent Platform. With the KStreams UI, teams get access to the KStreams experience, application discovery, ownership, client versions and health monitoring for Confluent platform via USM. Teams gain:

• Unified Observability: Bring KStreams visibility, application discovery, ownership, client versions and health monitoring for Confluent platform via USM. 

• Application health insights: Understand KStreams application health at a glance with full visibility into status, versions, and runtime behavior.

• Faster troubleshooting: Troubleshoot KStreams issues faster with a more focused operational experience for self-managed environments.

Client Monitoring on USM

Platform teams need better visibility into how producers and consumers are behaving across their Confluent Platform clusters. They would like to identify issues, optimize configurations, and understand which clients are connecting to the cluster.

Client monitoring on USM brings client-side observability to operators managing self-managed or hybrid deployment from Confluent Cloud. USM shows producer and consumer behavior, including throughput, connection details, and request activity collected directly from brokers without client-side instrumentation or extra agents.

RHEL 8 support for USM

USM Agent 1.2.0 adds RHEL 8 support, allowing teams to manage Confluent Platform through USM on RHEL 8 without changing their operating system baseline.

Migrate to KRaft with greater ease and confidence

Confluent Platform 8.3 includes several capabilities designed to help you migrate to Confluent for Kubernetes (CFK) KRaft with confidence. Whether you are migrating complex topologies, validating readiness earlier, or expanding quorum capacity with minimal disruption, Confluent Platform 8.3 provides you with the right tools and reduces operational overhead.

Support for ZooKeeper to KRaft Migration for 2.5 Datacenter (2.5DC) Deployments

Confluent Platform 8.3 simplifies your KRaft migrations for multi-region and 2.5DC deployments with a clearer supported path. For 2.5DC deployments, Confluent supports migrating the 0.5DC with a lite-mode (On K8s only), giving teams a clearer path to bring lightweight tiebreaker sites along as part of the broader migration. For multi-region clusters, CFK 3.3.0 and later simplifies migration by deriving the ZooKeeper endpoint for each region using Kafka, reducing manual configuration and making the move to KRaft more operationally straightforward. Together, these improvements create a more complete and supported migration path for distributed Kafka environments, helping teams reduce operational friction, plan with greater clarity, and carry the full topology forward to KRaft.

Support for pre-check utility for single-cluster KRaft migrations 

Migrate confidently by validating ZooKeeper metadata before single-cluster KRaft migrations with pre-check utility. This preflight check provides an additional pre-migration validation step, allowing teams to catch ZooKeeper metadata issues that standard configuration checks miss. Confluent recommends enabling this check for all single-cluster migrations, providing a proactive path to migration readiness. It is also operationally straightforward to adopt. It requires no additional RBAC permissions, is enabled with a single annotation, and automatically retries after restart once the underlying ZooKeeper issue is fixed.

Scale up the controller quorum, without restarting

Starting in CFK 3.3.0, teams can scale up controller quorum capacity in CFK without restarting brokers or existing controllers, making it easier to expand quorum capacity with minimal disruption. New controller pods start as observers, catch up with the leader, and can then be promoted to voters, creating a straightforward path to expand the quorum while maintaining cluster continuity. This gives platform teams a more flexible way to scale over time, improving operational agility while preserving uptime for the broker layer.

Structured, Centralized-Governed Data in Motion

Confluent Platform 8.3 introduces governance capabilities that enables structured, centralized governed data in motion and enforces data governance at the Gateway with zero client changes.

Schema IDs in Kafka headers

Schema IDs in Kafka headers decouples schema metadata from the payload, enabling simpler migrations from schemaless or custom formats to Schema Registry. While this capability was previously released on Confluent Cloud, Confluent is expanding support to Confluent Platform starting in version 8.3. With governed, structured data, you can fully leverage Confluent’s most powerful platform features, from Flink for real-time stream processing to Tableflow for unified lakehouse analytics, enabling you to govern streaming data at any scale. The capability provides:

• Zero-downtime schema adoption: Decouple metadata from the message body so teams can schematize existing streams without altering payload framing or home-grown consumers. 

• Broader ecosystem support: Already live on Confluent Cloud and now generally available in Confluent Platform 8.3, giving enterprises a single, standardized governance pattern across both cloud and self-managed environments.

• From "Dumb Pipes" to a Smart Data Plane: Moving metadata into headers turns raw event streams into a shared, intelligent data backbone. This makes it effortless to feed trustworthy, well-typed data into downstream AI, real-time Flink apps, and Tableflow lakehouse pipelines.

Centralized Data Governance Enforcement for CPC Gateway (Early Access)

With Centralized Data Governance Enforcement for CPC Gateway, you can enforce data quality, schema validation, and encryption centrally at the Gateway with zero client changes. This capability helps you gain:

• Enhanced Operational Control: Eliminate the need to coordinate updates across hundreds of application teams. Apply encryption, schema, or routing policies through a single gateway configuration point instead of managing library updates or manual configuration shifts across application teams.

• Improved Auditability and Compliance: Maintain a consistent protection layer across all data sources to support organizational auditing and meet stringent regulatory requirements, including for older and more complex applications.

• Simplified Client Management: Reduce the complexity and cost of maintaining hundreds of legacy client libraries across Java, C#, Python, and other languages to enforce data quality. Applications benefit from granular or full payload encryption without requiring changes to existing client application code.

Get Started Today

Download Confluent Platform 8.3 today. Confluent Platform is the only cloud-native and comprehensive platform for data in motion, built by the original co-creators of Apache Kafka. Before you upgrade to Confluent Platform 8.3, review the Confluent Platform upgrade guide and the Kafka 4.3 upgrade guide for detailed, step-by-step upgrade instructions, rolling upgrade considerations, and information about breaking changes and compatibility issues.

If you want to get Flink up and running, see the new Quick Start Guide that walks through installing CMF using Helm.

Ready to get started with Unified Stream Manager? The setup involves deploying the USM agent alongside your cluster, which can be automated using the Confluent for Kubernetes (CFK) operator or Ansible playbooks.

To learn more about CFK KRaft migrations, visit the Confluent for Kubernetes documentation page.

The preceding outlines our general product direction and is not a commitment to deliver any material, code, or functionality. The development, release, timing, and pricing of any features or functionality described may change. Customers should make their purchase decisions based on services, features,   and functions that are currently available.

Confluent and associated marks are trademarks or registered trademarks of Confluent, Inc.

Apache®, Apache Flink®, Flink®, Apache Kafka®, Kafka®,  and the Kafka and Flink logos are either registered trademarks or trademarks of the Apache Software Foundation in the United States and/or other countries. No endorsement by the Apache Software Foundation is implied by the use of these marks. All other trademarks are the property of their respective owners.