#509 Issue
June 29, 2026
Welcome to issue #508 June 22nd, 2026
News
Agents BigQuery Data Agent KitWhat’s new in data agents: Supercharging your AI workflows - Learn about new data agents and tools for business analysts, data scientists, and database admins to integrate with the Agentic Data Cloud.
Monitoring NetworkingCloud Network Insights: end-to-end observability for the Cross-Cloud Network - Gain end-to-end observability for Cross-Cloud Network. Cloud Network Insights provides proactive monitoring across hybrid and multicloud environments.
GCP Certification TPUUnlocking the Power of the TPU Stack: Introducing our new Developer Hub - Google has officially launched the TPU Developer Hub, a centralized educational resource designed to help model builders and developers maximize the performance of Google Cloud TPUs. The hub offers code-first resources, open-source recipes, and deep-dive documentation covering hardware architecture, software optimization, debugging, parallelism, and networking. These materials are tailored for both human developers and AI-assisted tools to streamline everything from large-scale training to low-latency inference workloads.
LookerIntroducing new Explores and Merge Queries in Looker - Looker Explore’s streamlined interface and integrated AI assistants make it easier for users to find answers in their data.
Agents AIAnnouncing the Agentic Resource Discovery specification - Google Cloud has introduced the Agentic Resource Discovery (ARD) specification, an open standard to enable AI agents to seamlessly find, verify, and securely connect with tools and capabilities across different organizations. This initiative aims to democratize AI resource discovery by providing a standardized layer for publishing and indexing agentic assets, fostering a more interoperable and scalable ecosystem.
Generative AI Model ArmorHow customer collaboration is shaping the future of GenAI security with Model Armor - Discover how Google enhanced Model Armor's security documentation for GenAI through customer collaboration and real-world developer insights.
Quadrant SecurityGoogle named a Leader in IDC MarketScape SIEM 2026 Vendor Assessment - We are proud to announce that Google has been named a Leader in the 2026 IDC MarketScape for worldwide SIEM platforms.
Articles, Tutorials
Infrastructure, Networking, Security, Kubernetes
Threat IntelligencePublic and Private Medical Community Targeted by China-Nexus Threat Actor Pursuing Artificial Intelligence, Cyber, Medical, and National Defense Research - UNC6508 leveraged Google Workspace compliance rules and REDCap vulnerabilities for intelligence collection.
CISOCloud CISO Perspectives: The 4 lessons that guided AI Threat Defense - Chris Betz, our new CISO, shares the 4 key lessons that helped us develop AI Threat Defense to fight AI with AI.
DevOps Google Kubernetes Engine KubernetesGKE Standby Buffers: Fast Scaling Without Idle Costs - Google Cloud's GKE Standby Buffers offer a cost-effective solution for fast scaling by pre-provisioning nodes in a suspended state, significantly reducing idle infrastructure expenses. This innovation allows for sub-second scheduling of unpredictable workloads by rapidly resuming nodes, ideal for complementing a hybrid strategy with active buffers for immediate spikes and standby for extended surges.
Google Kubernetes Engine RayScaling Ray Serve LLM on GKE: Performance without losing the developer experience - Through a partnership with Anyscale, Ray Serve LLM on GKE now offers 5x higher throughput and 8x lower latency for distributed inference.
DevOps GKE Autopilot SREActivepieces on GKE Autopilot: Six Engineering Disciplines for Production Workflow Automation - This article details deploying Activepieces, a workflow automation platform, on Google Cloud's GKE Autopilot for a robust production environment. It highlights how integrating six key engineering disciplines—Platform Engineering, GitOps, SRE, DevSecOps, CI/CD, and FinOps—creates a secure, reliable, and cost-effective solution. This "golden path" approach simplifies operations, enabling self-service provisioning of Activepieces instances with built-in security, observability, and automatic cost attribution.
InfrastructureIntroducing Brazos: Bringing liquid cooling to air-cooled data centers - The Google-developed Brazos is a rack-mounted, closed-loop liquid-to-air cooling system that you can deploy in an existing air-cooled environment.
InfrastructureThe Cloud Tetris: A Technical Deep Dive into GCP Sole-Tenant Node Usage Optimization - A practical guide to solving physical fragmentation and optimizing dedicated hardware compute costs in Google Cloud.
NetworkingReal-time IP capacity in Google Cloud subnets
Kubernetes LLMDeploying NVIDIA Nemotron-3 Ultra 550B, with B200 GPUs, vLLM on Google Kubernetes Engine — Football edition - This guide outlines the process of deploying the NVIDIA Nemotron-3 Ultra 550B Large Language Model on Google Kubernetes Engine, utilizing B200 GPUs and vLLM for high-performance inference. It details setting up a secure and private network architecture, complete with a custom Virtual Private Cloud and an internal load balancer. The deployment is then demonstrated through an interactive, football-themed CLI application that leverages the model's capabilities for dynamic simulation and commentary.
App Development, Serverless, Databases, DevOps
AI GCP ExperienceHow Siemens "slices the elephant," advancing agentic workflows for industrial software development - Siemens created Knowledge Fabric, an AI system for automating the software development lifecycle that reduced implementation effort.
Cloud SQL GCP ExperienceHow Atlas scales hundreds of merchant databases with Cloud SQL Enterprise Plus edition - See how Atlas scaled hundreds of merchant PostgreSQL databases and cut their overhead by migrating to Google Cloud SQL Enterprise Plus.
Gemini Generative AI PythonGenkit Middleware in Python: Retries, Fallbacks, Tool Approval - Genkit in Python offers a robust middleware system designed to enhance the reliability and resilience of Large Language Model (LLM) applications. It provides a consistent, composable pattern for addressing common issues like transient failures, model fallbacks, and securing human approval for tool execution. This allows developers to abstract away complexities such as retries and error handling, leading to more robust and production-ready AI applications.
Cloud Run Cloud Storage DevOpsHow We Served Multiple React Apps from a Single Domain Using GCP Load Balancer, Cloud Storage, and Cloud Run - This article details a robust Google Cloud Platform architecture for serving multiple React applications from a single domain. It demonstrates how to utilize a Global HTTP(S) Load Balancer, Cloud Storage for static hosting, and Cloud Run for dynamic metadata generation. This setup effectively manages SPA routing, deep linking, and social media previews while offering benefits like low maintenance and high scalability.
Cloud Filestore OracleMounting Google Cloud Filestore on Oracle AI Database@Google Cloud Exadata DB Nodes - A practical walkthrough for shared NFS storage with ExaDB-D.
Antigravity Go LLM TypescriptHow I learned Go in a Day with Antigravity 2.0 and How You Can Do the Same - Learn how to migrate a TypeScript CLI to Go in just one day using Antigravity 2.0. This guide covers architectural goals, AI-driven TDD, and building fast, secure software.
Cloud Run Docker LLM TerraformServerless GPU Inference: Deploy Any Hugging Face Model on Google Cloud Run
Big Data, Analytics, ML&AI
BigQueryOptimizing BigQuery performance for transactional use cases: Fine-grained DML - This article demonstrates how Google Cloud's BigQuery leverages its Fine-grained DML feature to optimize transactional use cases, such as real-time updates and deletions. Through comparative testing, it shows significant improvements in execution speed and reduced compute costs compared to traditional DML methods, by separating logical and physical data changes. This enhancement is crucial for efficiently handling high-volume, dynamic data operations within cloud-native data platforms.
BigQueryThe Hidden Powerhouse: Demystifying the BigQuery Storage Write API - The BigQuery Storage Write API is a powerful, gRPC-based engine that enables highly optimized and efficient streaming of data into BigQuery, often used implicitly by various Google Cloud services. This API significantly improves data ingestion performance over previous methods by leveraging Protocol Buffers and bidirectional streaming, offering flexible write modes to suit real-time, transactional, or buffered data requirements.
Airflow BigQuery Cloud StorageBackfilling 30 Million Files from GCS to BigQuery - This article details the complex journey of backfilling 30 million files from Google Cloud Storage into BigQuery, highlighting challenges often overlooked in data engineering. It describes the iterative process of building a robust pipeline, overcoming issues like API rate limits, data corruption, and system interruptions through strategies like idempotency and separating file discovery from ingestion.
Apache Iceberg BigQuery dbtBuilding a Serverless, Multi-Source Data Ingestion Framework on GCP (Snowflake & Databricks to BigQuery) - This article details a serverless, multi-source data ingestion framework developed on Google Cloud Platform, designed to consolidate data from Snowflake and Databricks into BigQuery. Leveraging GCP Workflows for orchestration and Dataproc Serverless for compute, this solution significantly reduces costs and operational overhead by eliminating persistent clusters.
AI BigQuery FinOpsHow AI made a rejected BigQuery cost optimisation worth shipping (loveholidays AI diary #8) - Loveholidays successfully implemented a BigQuery cost optimization strategy that was previously deemed too costly to pursue, by leveraging AI coding assistants. This AI-powered solution automates the intelligent switching between BigQuery's logical and physical storage billing models, leading to significant projected savings. The initiative demonstrates how a robust AI agent environment can enable the delivery of projects that were once considered unfeasible.
AI MCPA2UI + MCP Apps: Combining the best of declarative and custom agentic UIs - This post introduces three architectural patterns designed to integrate Model Context Protocol (MCP) Apps and Agent-to-User Interface (A2UI) to solve the tradeoff between highly custom iframe environments and native, declarative rendering. By combining these approaches, developers can serve native-feeling UIs directly over MCP servers, embed complex and stateful iframe apps securely inside declarative views, or inject generative UI components into legacy systems.
A2A AIHow A2A is Building a World of Collaborative Agents - Celebrating the first anniversary of the Agent-to-Agent (A2A) protocol, this blog post highlights how the framework enables autonomous AI agents to securely collaborate and hand off tasks without the rigidity of traditional APIs. By delegating complex workflows to specialized peer agents, A2A prevents context pollution, ensures data privacy, and simplifies application design through modularity.
Agents BigQuery GCP Experience Google Kubernetes EngineArchitecting a trusted agentic platform with graph technologies: A Yahoo case study - Using Google graph technologies, Yahoo's Seller Agent platform is a model for other industries striving to build a trusted agentic system of action.
Google Kubernetes Engine MCPBuild and Deploy a Remote MCP Server to GKE in 30 Minutes - Learn to build and deploy a secure remote MCP server on Google Kubernetes Engine (GKE) in 30 minutes. Streamline your AgenticAI workflows with standard protocols.
Agents AntigravityAgent Factory Recap: 100X engineering with AI agents in Google Antigravity 2.0 - Achieve 100X engineering with AI agents in Google Antigravity 2.0. Learn to scale impact, build skills, and automate workflows on this agent-first platform.
AI Open Knowledge FormatWhy Google’s Open Knowledge Format Matters for Developers and Content Creators?
Agents Open Knowledge FormatOpen Knowledge Format — portable digital map of your data as code - The Open Knowledge Format (OKF) is a proposed open standard for representing data as code, designed to create a portable digital map of information and address the challenge of scattered knowledge for AI agents. This article details OKF's minimal yet powerful approach to standardizing data descriptions.
Gemini Enterprise Agent Platform Workload Identity FederationGemini Enterprise Connectors: How Protocol & Pool Updates Affect Your Gemini Enterprise Connectors ? - This article explains how changes to your Workforce Identity Federation setup impact Gemini Enterprise Connectors, which securely access external data for search. While updating protocols or identity mappings within an existing identity pool is safe, changing the pool itself necessitates deleting and recreating connectors. This is because a new pool generates different user IDs, breaking access controls for previously indexed data and requiring a full re-crawl.
Data quality traffic lights - his article explains how Nordnet built a Data Quality Health Badge that combines incident detection, automated lineage tracking, anomaly detection, and dashboard trust signals to make data reliability visible directly inside Looker. It shows how surfacing real-time data health and impact analysis helps users, AI agents, and operational systems make better decisions while reducing silent failures, alert fatigue, and time spent tracing downstream issues.
Various
AI BusinessHow growing UK midsize businesses are building in the AI era - See how SMBs are doing more with AI, including 71% of AI adopters surveyed in the UK who say the technology helps them save time on routine tasks.
Agents AI BusinessFrom AI potential to agentic reality: Driving the UK’s next chapter - Organizations like HSBC, Ineffable Intelligence, Starling, Vodafone, and the UK and local governments are showcasing their transformation with Google Cloud at the London Summit.
Sovereign CloudChoice, compliance, and collaboration: Europe’s path to open digital sovereignty - Our Sovereign Cloud solutions are designed to meet Europe's tiered compliance requirements at every level.
StartupsScaling the Next Generation of Global Innovation: How Google Supports Top Startups Around the World - Discover how Google's accelerator program helps global startups scale with deep architectural guidance, expert engineering support, and a resilient global network.
Slides, Videos, Audio
GCP Bytes Podcast - #43 In this episode we discuss; C64, Bankys OS, GDG, Troy Beebe, Token Honeymoon, Telstra Automation, iSCSI vs NVMe/TCP, SpaceX Deal, Open Knowledge Format, Android Auto Disconnect, Unifi Bypass, Fable 5.
Releases
API Gateway - Update to the API Gateway runtime architecture The API Gateway runtime architecture is being updated to improve its integration with Google Cloud Platform and its services. This update does not affect existing API Gateway features. However, be aware of the following differences: Status code changes for gRPC API Gateways Error New status code Previous status code Quota exceeded ResourceExhausted Unavailable Invalid API key InvalidArgument InternalError For 4xx client-side quota failures, API Gateway will now reject requests (fail closed). This applies to both gRPC and OpenAPI API Gateways. If you experience any other differences in behavior due to this update, contact Google Cloud Customer Care. Note: Rollouts of this release to production instances might take up to 4 weeks to complete across all Google Cloud zones. Your instances might not be updated until the rollout is complete.
AlloyDB - The Database Insights remote Model Context Protocol (MCP) server now supports the following advanced query insights tools for AlloyDB for PostgreSQL: get_advanced_aggregated_query_stats get_advanced_aggregated_wait_event_stats get_advanced_time_series_query_stats get_advanced_time_series_wait_event_stats get_index_recommendations For more information, see Database Insights remote MCP server. AlloyDB integration with Knowledge Catalog is now generally available ( GA ). This integration provides a unified metadata view to simplify data governance and analysis. It includes near real-time synchronization and expanded metadata details, like primary and foreign keys. For more information, see Integrate AlloyDB with Knowledge Catalog.
Anti Money Laundering AI - New minor engine versions released for the commercial line of business within the v004.009 and v004.010 version lines. These versions extend support for the major engine version and include no significant changes compared to the previous minor versions.
Apigee Hybrid - v1.14.6 On June 16, 2026 we released an updated version of the Apigee hybrid software, v1.14.6. For information on upgrading, see Upgrading Apigee hybrid to version v1.14.6. For information on new installations, see The big picture. Note: This is a patch release: The container images used in patch releases are integrated with the Apigee hybrid Helm charts. Upgrading to a patch via the Helm chart automatically updates the images. No manual image changes are typically needed. For information on container image support in Apigee hybrid releases, see Apigee release process. Various security and CVE fixes are included in this release.
Backup and DR Service - Backup vault support for Cloud SQL instances encrypted with customer-managed encryption keys (CMEK) is generally available (GA), providing immutable and indelible storage with enforced retention. For more information, see Back up Cloud SQL instances to a backup vault.
BigQuery - Use Gemini Cloud Assist to analyze your SQL queries and receive recommendations to optimize query performance in BigQuery. This feature is available to customers who use BigQuery editions. This feature is in Preview. Support for configuring daily token quotas for BigQuery generative AI functions has been temporarily disabled. We are working to restore this feature as soon as possible. You can use Gemini Code Assist directly within the BigQuery Jobs explorer, Job details, Job history, and Capacity management pages to help you troubleshoot and analyze performance issues. For more information, see Troubleshoot job performance. This feature is in Preview. You can resize the width of table columns in BigQuery Studio for BigQuery listings such as datasets, repositories, job history, and connections. To resize a column, hover over the column divider and drag it to your preferred width. Use Gemini Cloud Assist to analyze your SQL queries and receive recommendations to optimize query performance in BigQuery. This feature is available to customers who use BigQuery editions. This feature is generally available (GA). Table Explorer behavior is moving to the Reference panel. This transition will occur in July 2026 or later. For more information, see Table Explorer. You can enable autonomous embedding generation on new or existing tables that you make with the CREATE TABLE or ALTER TABLE statements. When you do this, BigQuery maintains a column of embeddings on the table based on a source column. When you add or modify data in the source column, BigQuery automatically generates or updates the embedding column for that data. This feature is generally available (GA).
Bigtable - You can use the Bigtable Studio explorer to search for all resources except for authorized views and column families. For more information, see Manage your data using Bigtable Studio.
Billing - New filters and group-by options available in Cloud Billing Reports Cloud Billing has added two filters to the Billing Reports page to help you analyze and understand your costs: Products: Google Cloud Products consist of a group of SKUs (potentially from more than one Google Cloud Service ) that work together and are sold as a single service, sometimes referred to as a logical product family or a subscription service. Examples include Gemini Enterprise and Firebase App Hosting. Originating services: An Originating service is a Google Cloud service that causes usage in another service. For example, Google Kubernetes Engine (GKE) can cause usage in Compute Engine. In this use case, when you are viewing the Compute Engine usage and costs, GKE is an originating service when it causes usage in Compute Engine. You can also Group by the new filters, to summarize your costs by the dimension you select. Product: When you group by Product, the Report shows your costs and savings summarized by Product. Originating service > Service: When you group by Originating service > Service, the Report shows your costs and savings summarized by Originating service. In the report table, you can expand each row for an Originating service to see your costs summarized by each Service that is associated with the Originating service. Learn more about analyzing billing data and cost trends with Reports. Learn how to view Gemini Enterprise costs in Cloud Billing reports. CUD dashboard redesign available (preview) The redesigned CUD dashboard is available in the Billing section of the Google Cloud console. It provides a consolidated view of all your resource-based and spend-based CUDs in a single place. The new design improves usability and scalability, helping you find information faster. For more information, see View your commitments.
Blockchain Node Engine - Limited support for Blockchain Node Engine Starting June 15, 2026, Blockchain Node Engine will enter a period of limited support. New node creation in Blockchain Node Engine and provisioning of new Blockchain RPC endpoints will be disabled. Existing nodes and endpoints will continue to function and receive critical updates until the final shutdown date. We recommend migrating your workloads to our partner, Quicknode, to avoid service disruption. For more information, see the migration guide. Limited support for Blockchain Node Engine Starting June 15, 2026, Blockchain Node Engine will enter a period of limited support. New node creation in Blockchain Node Engine and provisioning of new Blockchain RPC endpoints will be disabled. Existing nodes and endpoints will continue to function and receive critical updates until the final shutdown date. We recommend migrating your workloads to our partner, QuickNode, to avoid service disruption. For more information, see the migration guide.
Chronicle - [Spotlight Feature] Ask Gemini Cloud Assist in Feed Management Google SecOps now provides Gemini Cloud Assist (GCA) directly within the Feed Management interface to help you with feed creation, setup, and general troubleshooting questions. A new Ask Gemini Cloud Assist button is now available in the Feed Management interface. You can click this button to open the Gemini Cloud Assist panel and ask questions to get guidance on: * Configuring and managing data feeds. * Understanding ingestion pre-requisites and setup steps for different log sources. * Resolving common setup issues. Note: Gemini Cloud Assist provides recommendations and answers to your questions, but does not perform configuration changes on your behalf. You must apply any recommended changes manually to your feeds. For more information, see Feed management overview. New Documentation changelogs Google SecOps is now releasing a monthly changelog to capture major documentation updates. For more information, refer to Documentation changelog. Auto-collapse setting for the query editor You can now configure the query editor to automatically collapse after you run a search, maximizing the screen space available for viewing your search results. By default, the query editor remains expanded. For more information, see Configure query editor behavior.
Chronicle SOAR - New Documentation changelogs Google SecOps is now releasing a monthly changelog to capture major documentation updates. For more information, refer to Documentation changelog. Release 6.3.89 is now available for all regions. Release 6.3.90 is being rolled out to the first phase of regions as listed here. This release contains internal and customer bug fixes. Scheduled Maintenance CloudSQL will undergo a scheduled minor upgrade.
Chronicle Security Operations - New Documentation changelogs Google SecOps is now releasing a monthly changelog to capture major documentation updates. For more information, refer to Documentation changelog. Auto-collapse setting for the query editor You can now configure the query editor to automatically collapse after you run a search, maximizing the screen space available for viewing your search results. By default, the query editor remains expanded. For more information, see Configure query editor behavior. Scheduled Maintenance CloudSQL will undergo a scheduled minor upgrade this Sunday, June 21, 2026.
Cloud Architecture Center - (New guide) Multi-tenant agentic AI system: A reference architecture to help you design a robust multi-tenant agentic AI system in Google Cloud.
Cloud Logging - The Cloud Logging API adds support for the ca regional endpoint. For a complete list of regional endpoints, see the REST reference pages.
Cloud Memorystore - Memorystore for Redis supports the General Availability of the following health issues: Expensive commands: resolve performance issues that are associated with using Redis commands that are resource-intensive (expensive). High resource utilization: resolve issues that are associated with instances not performing optimally. Maintenance policy not set: check whether users set maintenance windows for instances. If there's an optimal time slot for the maintenance windows when there's low traffic, then the health issue provides this slot.
Cloud SQL MySQL - Cloud SQL for MySQL now supports minor version 8.0.46. To upgrade your existing instance to the new minor version, see Upgrade the database minor version.
Cloud SQL Postgres - QueryData adds support for parameterized secure views (PSVs) to help secure applications that use natural language queries. For more information, see Secure and control access to application data. This feature is in Preview. Cloud SQL is integrated with Google AI Studio to help you build full-stack applications that use a Cloud SQL for PostgreSQL developer edition instance as the database. You can enter natural language prompts in the Google AI Studio to build applications backed by Cloud SQL and add features such as authentication, search, and persistent data storage. For more information, see Build vibe-coded applications using Google AI Studio and Cloud SQL. This feature is generally available ( GA ). The rollout of the following Cloud SQL for PostgreSQL minor version and extension upgrades is complete: Minor versions 14.22 is upgraded to 14.23. 15.17 is upgraded to 15.18. 16.13 is upgraded to 16.14. 17.9 is upgraded to 17.10. 18.3 is upgraded to 18.4. The new maintenance version is [PostgreSQL version].R20260319.07_04. To apply the new maintenance version, see Perform self-service maintenance.
Cloud Storage - When you create composite objects, you can delete the temporary source objects as part of the composition process.
Cloud Trace - You can collect, view, and analyze multimodal prompts and responses from your agentic applications that use the LangGraph or Agent Development Kit (ADK) frameworks. This feature is generally available (GA). Instrument generative AI applications Collect and view multimodal prompts and responses
Compute Engine - Preview: Before you create Spot VMs, you can view the following information for a specific machine type and location: You can view the real-time obtainability and estimated uptime. This information helps you maximize your chances of successfully creating Spot VMs, as well as help ensure that your workload starts and runs efficiently. You can view historical and current preemption rate and pricing. This information helps you compare and choose the configuration that best fits your workload needs and budget. For more information, see View the availability of Spot VMs and View the preemption rate and pricing for Spot VMs. For resource-based committed use discounts (CUDs), the default value of CUD scope for most Cloud Billing accounts has changed from Project to Billing account. If the CUD scope is set to Billing account, then resource-based CUDs from a commitment are shared across all projects in that account. If the CUD scope is set to Project, then resource-based CUDs from a commitment are available to only the project in which you purchased that commitment. Depending on the Cloud Billing account's creation date and the active commitments in that account, this change applies in the following way: Cloud Billing accounts created on or after June 16, 2026: The CUD scope is Billing account (CUD sharing enabled) by default. Cloud Billing accounts created before June 16, 2026: If the account has no active resource-based commitments on June 16, 2026, then the CUD scope has changed to Billing account (CUD sharing enabled). If the account has any active resource-based commitments on June 16, 2026, then the CUD scope remains unchanged and Google Cloud continues to use your existing configuration. For more information, see Share resource-based CUDs across projects.
Confidential VM - Support for the accelerator-optimized g4-standard-48 machine type for securely running AI and ML workloads is available in Preview, with the following specifications: 5th Generation AMD EPYC Turin processor AMD SEV 1 NVIDIA RTX PRO 6000 GPU
Contact Center AI Platform - Full information on the release page.
Dataflow - Dataflow has updated and expanded its pipeline update features for streaming jobs: Automated stop-and-replace updates: You can perform automated, declarative stop-and-replace updates to streaming jobs. Parallel updates with the same job name: When you perform automated parallel updates, you can use the same job name for the new replacement job. Auto-cancel draining jobs: When performing parallel or stop-and-replace updates, you can configure Dataflow to automatically cancel the old job if it does not finish draining after a timeout you specify. Update strategy configuration: You can explicitly choose between a parallel update ( update_strategy_parallel_job_update ) and a standard in-place update ( update_strategy_in_place_update ) while keeping all other configuration the same. Template upsert functionality: When launching pipelines from classic templates, flex templates, Terraform, or Config Connector, you can use the create_or_update_job experiment to enable automatic create-or-update (upsert) behavior. If an active job with the specified name already exists, it is updated. Otherwise, a new job is created. For more information, see Automated stop and replace, Automated parallel pipeline updates, and Automatic create or update (upsert) for templates. Dataflow now supports NVIDIA RTX Pro 6000 GPUs. You can use this GPU model to run your Apache Beam pipelines on Dataflow. RTX Pro 6000 GPUs are recommended for large, medium, and small model inference workloads. To configure your workers with this GPU model, set the accelerator type to nvidia-rtx-pro-6000. For more information, see Dataflow support for GPUs.
Dataproc - Managed Service for Apache Spark (formerly Dataproc on Compute Engine): Rollout of the new sub-minor versions without pre-configured channels will begin on June 22, 2026, delayed from the previously planned date of June 15, 2026 ETA.
IAM - You can use the error ID provided in permission error messages to help troubleshoot access. Error IDs provide context for the error, including the principal, resource, permission, and supported IAM conditions. This feature is available in Preview. For more information, see Permission error messages.
Memorystore for Redis Cluster - Memorystore for Redis Cluster supports the General Availability of the following health issues: Expensive commands: resolve performance issues that are associated with using Redis commands that are resource-intensive (expensive). High resource utilization: resolve issues that are associated with clusters not performing optimally. Maintenance policy not set: check whether users set maintenance windows for clusters. If there's an optimal time slot for the maintenance windows when there's low traffic, then the health issue provides this slot.
Secure Web Proxy - You can now use authorization policies to perform identity-based and content-based access control checks when processing outbound traffic requests through Secure Web Proxy. By configuring authorization policies, you can set rules for your workloads to access external destinations. You can also use these authorization policies to delegate complex authorization decisions to identity and content-scanning services like Service Extensions. This feature is supported in Preview. You can integrate frontend mutual TLS (mTLS) with Secure Web Proxy to boost the security of your applications and workloads. With this integration, you can use validated client identities in Secure Web Proxy authorization policies to enforce granular access control for outbound traffic. This feature is supported in Preview.
Security Command Center - Security Command Center External Exposure is available in Preview for the Security Command Center Premium tier. The service helps you manage and reduce your external attack surface through automated asset discovery, Google Cloud network exposure path validation, and active exploitability testing. For more information, see Detect exposed resources.
Service Extensions - Support for the ext_authz Envoy gRPC API protocol is now generally available (GA) for regional external Application Load Balancers and regional internal Application Load Balancers.
Service Mesh - The Envoy Compressor Filter is now GA in the rapid release channel. To ensure your EnvoyFilter compressor configuration is fully supported, see Modernize EnvoyFilter compressor configurations.