Newsletter from the Desk of Confluent Developer, July 16, 2026

Current San Francisco 2026 – Registration Opens in Early August

The biggest data streaming event of the year is heading to San Francisco this November and registration opens in early August.

Join thousands of data engineers, architects, developers, and technology leaders for two days of technical sessions, hands-on learning, product innovation, and real-world customer stories showcasing what's next in data streaming, Apache Kafka®, Apache Flink®, AI, and beyond.

This year, we're bringing the data streaming community together in the heart of San Francisco to explore the latest innovations, connect with industry experts, and discover how organizations are building intelligent, real-time applications.

Mark your calendar for registration opening in early August and be among the first to secure your spot at Current San Francisco 2026 HERE!

Queues for Kafka for Python Clients in Preview

The 2.15.0 release of the confluent-kafka-python client features a preview of support for KIP-932 - Queues for Kafka. This Preview release brings the shared consumption model to one of Kafka's most relevant languages, Python.

As a reminder, Queues for Kafka enables multiple consumers to cooperatively read from the same partition, with per-record acquire/acknowledge semantics and redelivery, so users can scale consumers beyond Kafka partition counts and distribute work like a task queue. This is a major step in bringing Queues for Kafka to the broader client ecosystem.

Python client (Preview): ShareConsumer

This preview includes new ShareConsumer and DeserializingShareConsumer clients with subscribe, batch poll, and close. Implicit and explicit acknowledgement modes, the ACCEPT / RELEASE / REJECT acknowledgement types, and delivery_count for poison-record detection. Synchronous and asynchronous commit plus an acknowledgement-commit callback, context-manager support, runtime SASL credential updates, and Schema Registry deserializers via DeserializingShareConsumer.

This Preview requires an Apache Kafka distribution that is on version 4.2.0 or higher.

⚠️This is a Preview: public interfaces may change before GA and it is not recommended for production use.

Here are some resources to help get you started with Queues for Kafka using Python

• Share consumer for the Python client
• API docs
• Python wheels
• Share Consumer core Apache docs
• Share Consumers on Confluent Cloud docs
• Share Consumers on Confluent Platform docs

What's Next?

We're working toward GA and broader feature parity with the Java share consumer, as well as share consumer support across the rest of the client family.

Announcing Early Access of Centralized Data Governance Enforcement for both CPC and CC

The 1.3 release of Confluent Private Cloud Gateway and Confluent Cloud Gateway enables Early Access of Centralized Data Governance Enforcement capabilities - Schema ID Enforcement, Deep Schema Validation, Field-Level Encryption and Payload Encryption.

This release is an important step forward in evolving Gateway beyond routing and connectivity into a stronger control point for Governance and Data quality.

What is new with Gateway 1.3?

• Schema ID Enforcement and Deep Schema Validation to help ensure only compliant, well-formed data flows through your pipelines.
• Field-level Encryption(GWFLE) and Payload Encryption(GWPE) to give teams more precise control over how sensitive data is protected in motion.
• Centralized Governance with Authoritative Enforcement Controls for platform operators — policies are defined once in Confluent Schema Registry and enforced through Gateway across all client applications.
• Support across Avro, JSON Schema, and Protobuf for all the Governance policies.

Additionally, there are new auth swap combinations like OAuth to OAuth and secret store integrations supporting CyberArk.

All these capabilities are available for both CPC Gateway and CC Gateway.

Why It Matters

• Enhanced Operational Control: Eliminates the organizational challenge of coordinating with hundreds of application teams for updates. Remove the need to ping hundreds of app owners for library updates or manual configuration shifts by applying encryption, schema, or routing policies through a single gateway configuration point.
• Improved Auditability and Compliance: Maintains a consistent protection layer across all data sources, which assists with organizational auditing and helps meet stringent regulatory mandates—even for the oldest, most complex applications.
• Simplified Client Management: Removes the complexity and cost associated with trying to update or maintain hundreds of legacy client libraries across Java, C#, Python, etc. just to enforce data quality. Applications benefit from granular or full payload encryption transparently with zero changes to existing client application code.

Check out these samples on GitHub for a hands-on demo.

Data Streaming News

• Apache Flink Agents 0.3.0 is Released: This includes Agent Skills - an emerging standard for packaging prompts, tools, and resources into self-contained capabilities that an agent can discover and load on demand. Long-Term Memory is now backed by Mem0 in both Python and Java, replacing the previous vector-store-based implementation. Flink Agents 0.3 introduces a declarative YAML API for describing agents in both Python and Java. There are also other features centered on reliability and observability.
• Kafka Rebalances: What’s Actually Happening Under the Hood: Ever wonder what’s really happening during a Kafka rebalance? Well, Mina Tafreshi has a piece to suss out the details - from group coordinators, to generation IDs, to what’s actually in the __consumer_offsets topic.
• Gilles Phillipart launches StreamSheets - a set of “cheatsheets” filled with common terminology for Apache Kafka, Apache Flink, and Apache Iceberg.
• Jack VanLightly starts a new series on Apache Kafka Performance using Dimster - starting with a producer experiment with linger.ms from version 3.7 to version 4.3 - as the default value changed from 0 to 5 as of the 4.0 release.

In-Person Meetups

• July 17th: Singapore
• July 18th: Bangalore
• July 18th: Hyderabad
• July 21st: Sydney
• July 27th: Melbourne
• July 29th: Jakarta

By the way…

We hope you enjoyed our curated assortment of resources! If you’d like to provide feedback, suggest ideas for content you’d like to see, or you want to submit your own resource for consideration, email us at devx_newsletter@confluent.io!