Google Cloud adds BigQuery streaming & Dataflow updates

Google Cloud has introduced updates across its data products, including new streaming query functions in BigQuery, a synthetic data generator for Managed Service for Kafka, and changes to Dataflow pipeline updates. The announcements span preview and generally available features across analytics and data infrastructure tools.

Three products are at the centre of the latest additions. BigQuery continuous queries now support stateful processing in preview, the Kafka synthetic data generator is now generally available, and Dataflow users can now stop and replace pipelines as part of a broader update process.

In BigQuery, the new stateful processing option extends continuous queries beyond stateless streaming operations. Users can now apply JOINs, aggregations, and windowing functions within streaming queries, allowing calculations over time windows such as a 30-minute average.

The change brings BigQuery's continuous query model closer to more complex stream processing tasks that often require separate tools. It also lets teams working with live data derive rolling metrics and combine datasets within the same streaming workflow.

Managed Service for Kafka has also gained a tool designed to simplify early testing of a new cluster. The synthetic data generator, now generally available, sends mock data into a Kafka cluster without requiring users to modify an existing client application or create a separate virtual machine.

According to Google, the tool can be activated in three clicks and begin sending data in less than two minutes. It is aimed at users who want to test a cluster or try product features before connecting production data sources.

Pipeline changes

Dataflow pipeline updates have also expanded. In addition to in-place updates, users can now stop and replace pipelines, and a parallel pipeline option has been added to move workloads from an old pipeline to a new one with less disruption.

The update also introduces a drain timeout setting intended to limit costs if processing becomes stuck. These Dataflow changes are now generally available.

Taken together, the announcements point to a focus on reducing operational friction in data systems. Rather than relying only on large platform changes, the latest releases emphasise practical controls for testing, migration, and live processing.

Another area of development is Google's Lakehouse product. Managed Lakehouse tables for Apache Iceberg are now in preview through the console, giving users Google-managed Apache Iceberg tables within the Lakehouse environment.

The feature is intended to reduce the need for duplicate data pipelines and synchronisation between BigQuery and open-source engines. The format supports read and write interoperability across multiple engines using a shared storage layer, while automated table management handles tasks such as compaction and partition tuning.

Wider push

The broader list of updates also shows continued investment in data access tools and analytics interfaces. Earlier additions included Google-built JDBC and ODBC drivers for BigQuery in preview, aimed at connecting Java applications and other software directly to the data warehouse.

Google has also been reshaping parts of its analytics and business intelligence portfolio. Data Studio is being reintroduced with a broader role that includes hosting BigQuery conversational agents and data applications built in Colab notebooks, while BigQuery Graph has entered preview as a graph analytics product.

Looker and BigQuery have also received a series of natural language and self-service updates. These include Conversational Analytics in BigQuery, Conversational Analytics for Looker Embedded environments, and self-service Explores in Looker.

Across databases, Google has disclosed further work tied to AI-related tooling and operational access. Announcements this year have included managed and remote MCP support for several Google Cloud databases, an enhanced Gemini assistant in BigQuery Studio, and new features in managed Airflow services.

Several customer references were also cited across the updates for Looker and related analytics products, including Telenor, Pet Circle, Fluent Commerce, Lighthouse Intelligence, Wego, ROLLER, Arrive, Audika, Carousell, Framebridge, GumGum, Intel, Overdose Digital, Ocean Network Express, Subskribe, and Promevo.

The latest product changes suggest Google is continuing to tie together analytics, streaming, orchestration, and database services inside its Data Cloud portfolio, with immediate attention on stream processing in BigQuery, test data generation for Kafka, and more flexible pipeline replacement in Dataflow.