The new confluent flink query command in the Confluent CLI runs a bounded Flink SQL statement as a snapshot query, waits for it to finish, and prints its results. Scripts and one-time queries no longer need to poll a statement handle for results. For more information, see Run a query.
Unified Stream Manager now supports registering and monitoring self-managed Connect clusters that run against Kafka clusters hosted on Confluent Cloud, in addition to Kafka clusters running on Confluent Platform. For more information, see Self-Managed Connectors on Confluent Cloud.
Unified Stream Manager now supports RBAC for your registered Confluent Platform clusters, with the UsmClusterAdmin, UsmOperator, and MetricsViewer roles providing administrative, read-only, and metrics-only access respectively. For more information, see Role-Based Access Control for Unified Stream Manager.
LATERAL SNAPSHOT joins in Confluent Cloud for Apache Flink are now generally available. Use a LATERAL SNAPSHOT join to enrich each row of a stream with the current version of a reference table, such as customer or product data. The join first loads the whole reference table, then joins each new row as soon as it arrives, so a reference table that rarely changes no longer stalls your results. It fa…
Streaming bootstrap in Confluent Cloud for Apache Flink is now available as an Open Preview feature. With streaming bootstrap, a continuous query builds its state from historical data in batch mode, then switches to processing live data as a streaming query. On AWS, streaming bootstrap can read historical records from Tableflow tables and recent records from Kafka. For more information, see Stream…
Confluent Cloud now supports the following client and Kafka Streams features from Apache Kafka® 4.4. To use these features, upgrade your Kafka Java Client and Kafka Streams applications to version 4.4 or later: Kafka Streams domain-specific language (DSL) operators can now opt in to headers-aware KeyValue, Window, and Session state stores, preserving record headers through state store reads and wr…
Confluent Cloud for Apache Flink now supports managed AI models in compute pools that run on AWS, Azure, and Google Cloud. Confluent manages the models for you, so you can run AI model inference without creating a connection to a model provider. For more information, see Run a Managed AI Model.
The fully managed HTTP Source V2 and HTTP Sink V2 connectors now support manual configuration without an OpenAPI specification file. To configure a connector manually, turn off the Use Open API specs toggle, and then enter the API base URL and endpoint paths. For more information, see HTTP Source V2 and HTTP Sink V2.
You can now sign in to the managed MCP servers, including the Real-Time Context Engine MCP server, with your Confluent Cloud account through OAuth instead of an API key. MCP tools act with your own permissions. For more information, see Confluent Cloud account sign-in (OAuth).
Tableflow bills for index-related processing when you use the upsert write mode. For more information, see Billing with Tableflow in Confluent Cloud.
Secret manager integration with Confluent Cloud now extends to the fully managed Snowflake Source connector. For more information, see Create a secret manager integration in Confluent Cloud.
You can now monitor the health of your KRaft controllers in USM using the Confluent Cloud Console. The Brokers tab for a registered Confluent Platform cluster now shows the total controller count, the active controller ID, whether the cluster has a quorum, and a graph of active and standby controllers over time. KRaft controller monitoring requires a supported Confluent Platform version running in…
Two Debezium Single Message Transformations (SMTs) are now available for fully managed Debezium change data capture (CDC) Source connectors in Confluent Cloud: ExtractChangedRecordState: Adds message headers listing fields changed and unchanged by an UPDATE operation. This allows downstream consumers to detect field-level changes without comparing full before and after payload structures. For more…
The cross-region and cross-cloud connectivity support has now been extended to the fully managed Google BigQuery V2 Sink and MongoDB Atlas Sink connectors for Confluent Cloud. You no longer need to contact Confluent Support to connect to a destination in a different region or cloud provider than your Confluent Cloud cluster.
Updated Confluent Kafka clients powered by librdkafka v2.16.0 are now available for Python Client, JavaScript Client, .NET Client, Go Client, and C/C++ Client. This release delivers: Free threading support for Python Client IBM Z s390x platform support for Python Client, JavaScript Client, and .NET Client Topic Subject Association admin APIs For more information, see librdkafka and derived client…
Freight clusters now support key-based compacted storage, the idempotent producer, transactions, and the DELETE_RECORDS and REPLICA_STATUS operations. For more information, see Freight cluster considerations.
The fully managed Marketo Sink connector is now generally available for your Apache Kafka® clusters on AWS, Azure, and Google Cloud. Configure it to stream records from Kafka into Adobe Marketo Engage to upsert leads, log custom activities, upsert custom objects, and manage static list membership. To explore its full range of features and get started, see Marketo Sink Connector for Confluent Cloud…
The fully managed Azure Data Explorer (Kusto) Database Sink connector for Confluent Cloud is now generally available for your Apache Kafka® clusters on AWS, Azure, and Google Cloud. The connector streams events from Apache Kafka® topics into Azure Data Explorer (Kusto) database tables so you can query and analyze the data using the Kusto Query Language (KQL). For more information, see Azure Data E…
mTLS authentication is now generally available on Azure Enterprise clusters. For details, see Configure Mutual TLS (mTLS) Authentication.
Tableflow is now generally available on Google Cloud. You can now expose Apache Kafka® topics as Apache Iceberg™ tables stored in Confluent Managed Storage or in your own Google Cloud Storage buckets. For more information, see Store Tableflow data in Google Cloud Storage.
The fully managed Oracle XStream CDC Source connector now captures Oracle RAW and XMLType columns. Custom snapshot SELECT statements and ad-hoc blocking snapshot filters that query an XMLType column must explicitly serialize it to a CLOB using XMLSERIALIZE. For more information, see Large object (LOB) type and XMLType handling and Supported Data Types.
You can now delete consumer groups in the empty state through Confluent Cloud Console. For consumers that use assign(), the group appears empty and accepts deletion requests, but the consumer can still consume. For more information, see Use the Confluent Cloud Console to monitor Kafka consumer lag.
Confluent Cloud audit logs now record auditable events for System for Cross-domain Identity Management (SCIM) provisioning operations, including enabling SCIM for an organization, managing SCIM tokens, and creating, updating, or deleting users through a SCIM client. For more information, see SCIM Auditable Event Methods on Confluent Cloud.
The Advanced Security add-on is now available on Confluent Cloud as a Limited Availability release. This organization-level add-on bundles advanced security capabilities, including System for Cross-domain Identity Management (SCIM), self-managed encryption keys (BYOK), and higher limits for select security-related resources such as group mappings, identity pools, and RBAC role bindings. The add-on…
The Flink Table API for Confluent Cloud for Apache Flink is updated to Java plugin version 2.3-6 and Python package version 2.3.3 on Flink 2.3.0, adding the following features: AI functions: The AI_SENTIMENT and AI_DETECT_PII scalar functions and the AI_FORECAST and AI_DETECT_ANOMALIES aggregate functions are now supported through ConfluentMLExpressions. Changelog functions: The FROM_CHANGELOG and…
Tableflow now supports topics that use RecordNameStrategy or TopicRecordNameStrategy as a Limited Availability feature. You can materialize topics with multiple event types into a single table. To request access, contact Confluent Support. For more information, see Using RecordNameStrategy or TopicRecordNameStrategy.
The fully managed Oracle XStream CDC Source connector includes the following updates: Optional SELECT_CATALOG_ROLE: The connector no longer requires granting SELECT_CATALOG_ROLE to the connect user. You can instead grant SELECT on an explicit list of catalog views. For more information, see Configure database users. Active Data Guard standby support: For cascading downstream capture, the connector…
The logical cluster ID OAuth extension, for example, logicalCluster or extension_logicalCluster, is no longer required for Kafka clients connecting to the following cluster types, on any cloud provider or networking type: Dedicated clusters Freight clusters Enterprise clusters with an lkc-based bootstrap URL For more information, see OAuth Configuration Reference for Confluent Cloud Clients.
User-defined functions (UDFs) in Confluent Cloud for Apache Flink® are now available in one additional AWS region and two additional Azure regions: AWS: ap-southeast-7 (Bangkok, Thailand) Azure: italynorth (Milan, Italy), mexicocentral (Querétaro, Mexico) For the region-by-region capability matrix across all clouds, see UDF regional availability.
Custom SMTs on fully managed connectors in Confluent Cloud can now make external network calls (egress) as a Limited Availability feature. Enabling external egress allows your Custom SMT to connect to endpoints outside Confluent Cloud, such as key management services (KMS), secret stores, or data enrichment APIs. For more information, see Enable External Network Calls from Custom SMTs.