Google Cloud Platform Newsletter

Check Archive for older issues

News

Data Studio returns as new home for Data Cloud assets - Data Studio, formerly Looker Studio, is playing a significant role serving Google Data Cloud content, on top of its data visualization capabilities.

QueryData helps agents turn natural language into queries for AlloyDB, Cloud SQL and Spanner - Translate natural language into near-100% accurate database queries with QueryData - a tool for Data Agents.

Openness without compromises for your Apache Iceberg lakehouse - Apache Iceberg tables in Google-managed Iceberg REST Catalog provides read and write interoperability between BigQuery and Iceberg-compatible engines.

New GKE Cloud Storage FUSE Profiles take the guesswork out of configuring AI storage - New GKE Cloud Storage FUSE Profiles automate performance tuning and accelerate data access for your AI/ML workloads with minimal overhead.

Raising the security baseline: Essential AI and cloud security now on by default - To support the next generation of AI innovators, we are offering on by default essential AI security and cloud security in Security Command Center Standard.

Claude Mythos Preview: Available in private preview on Vertex AI - Claude Mythos Preview, Anthropic’s newest and most powerful model, is now available in private preview to a select group of Google Cloud customers.

See beyond the IP and secure URLs with Google Cloud NGFW - Announcing domain filtering with a wildcard capability in Cloud NGFW Enterprise, providing increased security and granular policy controls.

Build music generation into your apps with Lyria 3 models on Vertex AI - Lyria 3, Google's family of music generation models, is now available on Vertex AI in public preview. With Lyria 3, you can generate high-quality, high-fidelity stereo audio from text prompts and from images with vocal support.

Google Cloud named a Leader in The Forrester Wave™: Sovereign Cloud Platforms, Q2 2026 - Google Cloud has been named a Leader in The Forrester Wave™: Sovereign Cloud Platforms, Q2 2026, validating our portfolio of choice approach.

Introducing Looker self-service Explores for faster ad-hoc analysis - Self-service Explores in Looker let you quickly and effectively analyze data governed by the Looker semantic layer.

Conversational Analytics now available for Looker Embedded environments - Conversational analytics support in Looker Embedded environments extends the natural language experience to more users from more surfaces.

Articles, Tutorials

Infrastructure, Networking, Security, Kubernetes

AI infrastructure efficiency: Ironwood TPUs deliver 3.7x carbon efficiency gains - The seventh-generation Ironwood TPU achieved an approximately 3.7x improvement in Compute Carbon Intensity (CCI) compared to TPU v5p architecture.

Migrating to Google Cloud’s Application Load Balancer: A practical guide - Strategic and 4-phase approach for migrating on-premises application load balancer to Google cloud application load balancer using declarative features, and service extensions.

Experimenting with GPUs: GKE managed DRANET and Inference Gateway AI Deployment - Unleash next-level AI performance! Dive into the step-by-step guide for deploying the Deepseek LLM on GKE, harnessing managed DRANET, NVIDIA B200 GPUs, and the GKE Inference Gateway for lightning-fast, high-scale inference.

GKE's Noisy Neighbor Problem Can Be Invisible in Metrics Explorer - This technical overview examines how sampling intervals in Cloud Monitoring can mask CPU saturation on Google Kubernetes Engine (GKE) nodes.

Guardrails at the gateway: Securing AI inference on GKE with Model Armor - Here’s how to secure AI inference on Google Kubernetes Engine with Model Armor and high-performance storage.

6 Nuances of Binary Authorization That Are Hard to Find in the Docs - Binary Authorization provides a software supply chain security framework for ensuring that only trusted container images are deployed on Google Cloud. While the service is essential for establishing a secure-by-default posture, several architectural and operational nuances are critical for successful implementation in production environments.

Engineering a Perimeter: My Journey Automating Threat Intelligence at Scale - This article details an engineering project to automate threat intelligence at scale, focusing on blocking malicious IP addresses using Google Cloud. By leveraging services like Cloud Armor and Cloud Run, the system transformed reactive security into a proactive, "set and forget" defense, significantly enhancing infrastructure protection and operational efficiency.

A developer’s guide to architecting reliable GPU infrastructure at scale - Understand the core challenges behind reliable GPU-based infrastructure on Google Cloud, and find resources to dive deeper.

Setting Up a GCP Landing Zone for Organizations with Strict Regulatory Requirements - Setting up Google Cloud Platform (GCP) landing zones for organizations with strict regulatory requirements can be a complex task. This article introduces Merlin Studio, a tool designed to simplify this process by guiding users through defining business needs, configuring parameters, and generating compliant Cloud Foundation Fabric files and documentation.

I got tired of reading GCP’s IAM permissions page, so I built something better - Navigating Google Cloud IAM permissions can be challenging due to extensive documentation and a lack of effective search and comparison tools. A new web application, Permiso, addresses this by providing a searchable, filterable interface for GCP predefined IAM roles and their associated permissions. This tool helps users efficiently find, compare, and get recommendations for roles, significantly streamlining the process of managing cloud access.

Google Cloud Heroes #1: Why PAM alone isn’t enough to save your IAM? - A deadly duo: Standing privileges and Least Privilege.

App Development, Serverless, Databases, DevOps

A Developer’s Guide to the AppOptimize API: Natively Supported vs. Custom Code - The Google Cloud AppOptimize API empowers developers to programmatically retrieve cost and usage data for specific projects and applications, significantly streamlining cloud infrastructure cost management.

Near-100% Accurate Data for your Agent with Comprehensive Context Engineering - Learn why and how comprehensive context engineering ensures your agents are well-equipped to take action with near-100% accuracy.

How Estée Lauder Companies uses Cloud Run worker pools for its pull-based agentic workloads - With Cloud Run worker pools, now GA, Estée Lauder Companies has a platform for its pull-based, stateful agentic workloads.

Fine-Tuning Gemma 4 with Cloud Run Jobs: Serverless GPUs (NVIDIA RTX 6000 Pro) for pet breed classification - This guide explores fine-tuning Google's advanced Gemma 4 model for pet breed classification using Cloud Run Jobs and serverless NVIDIA RTX 6000 Pro GPUs. It details the essential code modifications required to adapt Gemma 4's multimodal architecture, overcoming challenges like activation clipping and memory management with QLoRA. The process demonstrates significant accuracy improvements, offering an efficient and scalable solution for training large language models.

Gcon - GCon is a powerful, keyboard-driven terminal UI (TUI) for managing Google Cloud Platform resources. It provides a fast, intuitive alternative to the GCP Console and gcloud CLI, with features like async operations, fuzzy search, and real-time resource monitoring.

Big Data, Analytics, ML&AI

Ultimate prompting guide for Lyria 3 models - Lyria 3 and Lyria 3 Pro are our most advanced music generation models, designed to give you granular control over vocals, instrumentation, and arrangement. Our guide shares exactly what we learned and how you can get the best results.

Accelerating data curation with Google Data Cloud - Learn about curation accelerators in Google Cloud to automate organizing, cleaning, and enriching raw data to transform it into AI-ready data assets.

How a leading consumer insight brand uses Dataproc to hyper-personalise faster - RVU built a platform that powers hundreds of automated personalization campaigns, optimized with billions of data points from across all its brands.

How SAP Concur automates expense reporting with agentic AI - Standard expense automation can see what’s on a receipt but not what’s missing. SAP Concur saw the emergence of AI agents as an opportunity to create systems that could reason, decide, and act.

Behind the Analysis with Google Cloud and Team USA: Architecting AI infrastructure for U.S. Winter Olympians - The AI pose estimation model transforms a single 2D video into a complete 3D biomechanical analysis, plotting 63 joints in a localized coordinate system — providing athletes a revolutionary competitive edge.

Under one roof: Rightmove reinvents property search with unified data - By migrating from on-prem databases to Google Cloud, Rightmove was able to unlock smarter, faster, more personalized experiences for its users and partners, so they could find the property more efficiently.

The Practical Guide to BigQuery Graph: Resources, Codelabs, and GQL Examples - Google Cloud's new BigQuery Graph allows for native modeling and analysis of complex relationships directly within BigQuery, leveraging Graph Query Language (GQL) and SQL/PGQ standards. The article provides a comprehensive guide with resources, hands-on codelabs, and practical GQL examples to help users master this powerful capability.

AI Functions in BigQuery Dataframes - Google Cloud's BigFrames introduces AI functions to bridge the gap between traditional dataframes and Generative AI workflows. This enables easy integration of Gemini-powered insights, moving beyond manual looping for complex AI analyses.

The Discipline of Data: A Complete Guide to Looker and LookML - From first principles to production-grade practice — for engineers who know SQL and are ready to build systems that outlast them.

Deep Dive into Google Cloud Pub/Sub Single Message Transforms and AI Inference - Learn how to use Pub/Sub Single Message Transforms (SMTs) to process data inline.

Building Event-Driven Data Agents with BigQuery, Pub/Sub, and ADK - Build real-time, autonomous AI agents to detect and resolve anomalies on Google Cloud. Combine BigQuery continuous queries, Pub/Sub routing, and Vertex AI Agent Engine (ADK) to instantly act on data.

How to run evals for Conversational Analytics agents - Moving an AI agent from a prototype to a production-ready tool requires rigorous, repeatable testing. Prism is an open-source evaluation tool for Conversational Analytics in the BigQuery UI and API, as well as the Looker API.

Create Expert Content: Local Testing of a Multi-Agent System with Memory - Learn how to locally test your multi-agent system, Dev Signal, ensuring its core reasoning and tool logic are properly integrated before cloud deployment. This guide details how to validate the full content creation lifecycle—from discovery to drafting—and confirms successful retrieval of long-term user preferences from the cloud-based Vertex AI memory bank.

TorchTPU: Running PyTorch Natively on TPUs at Google Scale - TorchTPU is a new engineering stack designed to provide a native, high-performance experience for running PyTorch workloads on Google’s TPU infrastructure with minimal code changes. It features an "Eager First" approach with multiple execution modes and utilizes the XLA compiler to optimize distributed training across massive clusters. Moving into 2026, the project aims to further reduce compilation overhead and expand support for dynamic shapes and custom kernels to ensure seamless scalability for the next generation of AI.

Is MCP dead? 2025 vs 2026 at GCP Next - Analysis of Google Cloud Next session catalogs reveals a substantial increase in "MCP" related content from 2025 to 2026, signaling Google's shift towards an AI agent execution stack.

vLLM: Production AI Inference on GKE with OpenWebUI, Gateway API, and TLS - This article outlines how to build a production-grade AI inference platform on Google Kubernetes Engine (GKE) Autopilot. It details using vLLM for high-throughput, low-latency LLM serving with features like continuous batching and PagedAttention, connected to OpenWebUI for a user-friendly interface.

Building a Healthcare Recommender with Keras, Two Towers and Google Cloud - This article details the creation of a healthcare recommender app that translates plain language symptoms into likely diseases, along with suggested medications, diets, and workouts. The system leverages Keras Recommenders with a Two-Tower architecture for semantic matching, ScaNN for efficient search, and is deployed via a Flask API on Google Cloud Run. It highlights how this architecture enables seamless scaling to Vertex AI Vector Search for production-grade performance and maintainability.

Various

Accelerating innovation and impact across the public sector - Discover why Gartner named Google a "Company to Beat" for agentic AI. Learn how Gemini for Government scales mission-critical workflows with secure AI.

Slides, Videos, Audio

Security Podcast - #271 Can AI-Native MDR Actually Fix Your Broken SOC Workflows or Just Automate the Mess?

GCP Bytes Podcast - #38 In this episode we discuss; Google Billing, Google Plans, Google Home Assist, OpenClaw, Youtube Premium, GitHub, GDG, Apple Hire, Alphabet Has Dropped, Cloud Spending, APRA, China and NVIDIA, Google and Redshift, Mythos, TurboQuant, Gemma4, Claude Code Source Leak.

Releases

Cloud Storage - Feature: You can now use Storage batch operations to update object contexts for multiple objects in a single job. You can clear all existing contexts from the specified objects, remove contexts with specific keys, or update and insert new context key-value pairs. For more information, see Create and manage batch operation jobs. Feature: Object contexts are now generally available. You can attach key-value pairs to your objects to categorize, track, and search your data. Object contexts are preserved by default during copy, rewrite, and compose operations. You can help control this behavior by using the dropContextGroups JSON API parameter or by providing new contexts in the request. Feature: You can delete up to 1,000 objects in a single request by using the Cloud Storage multi-object delete XML API. If you use Amazon S3-compatible tools or libraries, you can point your request to the Cloud Storage endpoint to use this feature with your existing workflows. For more information, see Delete objects and Delete multiple objects.

Cloud Architecture Center - Feature: (New guide) Multimodal GraphRAG resource orchestration: A high-level architecture to build and deploy a multi-agent AI system that consolidates fragmented multimodal data into a searchable knowledge graph. Feature: (New guide) Guide technical workflows with bidirectional multimodal streaming: A high-level architecture to build and deploy a multi-agent AI system that provides technical guidance and automated safety monitoring through a continuous, bidirectional stream of multimodal data. Feature: (New guide) Orchestrate security operations workflows: A high-level architecture to build a multi-agent AI system that orchestrates complex investigation and triage processes in a security operations center (SOC).

Cloud SQL SQL Server - Feature: Cloud SQL for SQL Server integration with Microsoft Entra ID ( GA ) provides centralized identity and access management (IAM) for your databases using your existing Microsoft Entra ID tenant. Feature: Cloud SQL for SQL Server now supports SQL Server 2025 ( GA ): SQL Server 2025 Standard SQL Server 2025 Enterprise SQL Server 2025 Express For more information, see Database versions and version policies and Choose a machine series. Feature: If the storage capacity of a Cloud SQL instance is larger than your application needs, then you can manually reduce, or shrink, your storage capacity to a smaller size. Depending on underlying disk size, storage shrink operations might incur considerable downtime. If your instance requires limited downtime, rather than using storage shrink capabilities, we recommend migrating your data to a new, smaller instance using Database Migration Service. For more information, see About storage shrink.

Cloud SQL - Feature: The QueryData tool lets you to query with the data in your database using conversational language and build data agents. For more information, see QueryData tool overview. This feature is available in ( Preview ). The preview release increases the accuracy of SQL generation with value search queries which match values and their context within a database. Value search queries trigger automatically. Feature: The QueryData tool lets you to query the data in your database using conversational language and build data agents. For more information, see QueryData tool overview. This feature is available in ( Preview ). The preview release increases the accuracy of SQL generation with value search queries which match values and their context within a database. Value search queries trigger automatically. Feature: If the storage capacity of a Cloud SQL instance is larger than your application needs, then you can manually reduce, or shrink, your storage capacity to a smaller size. Depending on underlying disk size, storage shrink operations might incur considerable downtime. If your instance requires limited downtime, rather than using storage shrink capabilities, we recommend migrating your data to a new, smaller instance using Database Migration Service. For more information, see About storage shrink. Feature: Context sets (previously known as data agents) enable you to interact with the data in your database using conversational language and are used by tools like QueryData to build conversation agents. For more information, see Context sets overview. This feature is available in ( Preview ). The preview release increases the accuracy of SQL generation with value search queries which match values and their context within a database. Value search queries trigger automatically.

Cloud Spanner - Feature: The QueryData tool lets you to query the data in your database using conversational language and build data agents. For more information, see QueryData tool overview. This feature is available in ( Preview ). The preview release increases the accuracy of SQL generation with value search queries which match values and their context within a database. Value search queries trigger automatically. Feature: Context sets (previously known as data agents) enable you to interact with the data in your database using conversational language and are used by tools like QueryData to build conversation agents. For more information, see Context sets overview. This feature is available in ( Preview ). The preview release increases the accuracy of SQL generation with value search queries which match values and their context within a database. Value search queries trigger automatically.

AlloyDB - Feature: Context sets (previously known as data agents) enable you to interact with the data in your database using conversational language and are used by tools like QueryData to build conversation agents. For more information, see Context sets overview. This feature is available in ( Preview ). The preview release increases the accuracy of SQL generation with value search queries which match values and their context within a database. Value search queries trigger automatically. Context sets also add support for Parameterized secure views (PSVs) to help secure applications that use natural language queries generated using QueryData. Feature: The QueryData tool lets you to query the data in your database using conversational language and build data agents. For more information, see QueryData tool overview. This feature is available in ( Preview ). The preview release increases the accuracy of SQL generation with value search queries which match values and their context within a database. Value search queries trigger automatically. QueryData also 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 using parameterized secure views. Feature: The QueryData tool lets you to query with the data in your database using conversational language and build data agents. For more information, see QueryData tool overview. This feature is available in ( Preview ). The preview release increases the accuracy of SQL generation with value search queries which match values and their context within a database. Value search queries trigger automatically. QueryData also 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 using parameterized secure views.

Chronicle - Full details on release page.

Gemini - Deprecated: Custom IAM roles permission update for Gemini Cloud Assist On April 8, 2026, Gemini Cloud Assist is replacing the cloudaicompanion.instances.completeTask IAM permission with geminicloudassist.agents.invoke. Updates to standard IAM roles will be done automatically, but if you have access to Gemini Cloud Assist through a custom IAM role, you must update the role before April 8, 2026 to ensure continued access. For more information, see the deprecated IAM permissions page. Breaking: Custom IAM roles permission update for Gemini Cloud Assist Gemini Cloud Assist has replaced the cloudaicompanion.instances.completeTask IAM permission with geminicloudassist.agents.invoke. If you have access to Gemini Cloud Assist through a custom IAM role, you must update the role to continue having access. For more information, see the deprecated IAM permissions page. Fixed: Agent mode logs now attributed to Gemini Code Assist VS Code Gemini Code Assist 2.77.1 now attributes agent mode logs to Gemini Code Assist. In previous versions, agent mode logs are being attributed to Gemini CLI instead of Gemini Code Assist. This discrepancy is resolved in the latest release, and we recommend that you update to version 2.77.1 or higher to ensure your usage metrics are correctly reported.

Apigee API Hub - Feature: Agent Registry integration support for MCP metadata (Preview) API hub now includes a managed integration with Agent Registry to automatically synchronize Model Context Protocol (MCP) servers and tools metadata. This feature enables AI agents to discover and interact with the APIs registered in your hub without manual configuration. This feature is in Public Preview. For more information, see Manage Agent Registry integration.

Cloud PubSub - Feature: Pub/Sub now offers the AI Inference Single Method Transform (SMT). This SMT lets you get inferences on Pub/Sub messages from Vertex AI models. The model's inferences are added to each message, making them available for downstream processing along with the original message data. The change is being rolled out in a phased manner over the rest of the week. For more information, see AI Inference SMT. This feature is generally available.

Google Cloud Armor - Feature: Cloud Armor preconfigured rules support ModSecurity Core Rule Set (CRS) 4.22 as a rule source. For more information, see Tuning Google Cloud Armor WAF rules. This feature is available in Preview.

Cloud Memorystore - Feature: You can use the Memorystore for Redis remote MCP server. This server lets you connect to Memorystore for Redis instances from LLMs, AI applications, and AI-enabled development platforms. This feature is available in Preview.

Memorystore for Redis Cluster - Feature: You can use the Memorystore for Redis Cluster remote MCP server. This server lets you connect to clusters in Memorystore for Redis Cluster from LLMs, AI applications, and AI-enabled development platforms. This feature is available in Preview.

BigQuery - Feature: You can now use the AI.AGG function to semantically aggregate unstructured input data based on natural language instructions. This feature is in Preview. Feature: You can now use a custom organization policy to allow or deny specific operations on these BigQuery resources: tables, data policies, and row access policies. This feature is in preview. Feature: The BigQuery Data Transfer Service now supports incremental data transfers for the following data source connectors: MySQL Oracle PostgreSQL ServiceNow These features are supported in Preview. Feature: You can now use the built-in text embedding model embeddinggemma-300m in the AI.EMBED and AI.SIMILARITY functions. This model uses your BigQuery slots to generate embeddings at scale. This feature is in Preview. Feature: You can now use the @@session_id system variable with SQL user-defined functions, table functions, and logical views. This feature is generally available (GA). Feature: The BigQuery Data Transfer Service now supports incremental data transfers when transferring data from Microsoft SQL Server to BigQuery. This feature is supported in Preview. Feature: The BigQuery Data Transfer Service can now transfer data from Snowflake to BigQuery. This feature is generally available (GA). Feature: You can now use BigQuery Graph to model your data as a graph and perform analysis on a large scale. Create a graph directly from tables that store entities and relationships between entities. You don't need to modify your existing workflows or replicate your data to use it in graph queries. Use Graph Query Language (GQL) to find complex, hidden relationships between data points that would be challenging to find using SQL. Visualize your graph schema and graph query results in a notebook. This feature is in Preview. Feature: You can now use stateful operations in continuous queries, which let you perform complex analysis by retaining information across multiple rows or time intervals using JOIN s and windowing aggregations. This feature is in Preview. Feature: SQL cells in BigQuery notebooks are now generally available (GA).

Container Registry - Feature: You can manually prewarm images in Artifact Registry to reduce the cold-start latency for deployments. This feature is only available using the API.

Artifact Registry - Feature: You can manually prewarm images in Artifact Registry to reduce the cold-start latency for deployments. This feature is only available using the API.

Dataflow - Feature: When you use min_ram or cpu_count resource hints for pipeline steps that don't require accelerators, Auto VM Selection (Instance Flexibility) is enabled automatically. With Auto VM Selection, workers are provisioned from a curated list of machine types that meet your RAM and CPU requirements. For more information, see Auto VM Selection for worker machine types.

IAM - Feature: Organization Policy Service custom constraints are available for managed workload identity and Workload Identity Federation. You can use custom constraints to control how managed workload identity and Workload Identity Federation are used in your organization. For more information, see Custom organization policy constraints for managed workload identity and Custom organization policy constraints for Workload Identity Federation.

Database Migration Service - Feature: You can use the Database Migration Service MCP server to enable agents and AI applications to view and manage running migration jobs. This feature is in Preview.

Confidential VM - Feature: Live migration is generally available (GA) on Confidential VM instances that meet the following configuration criteria: A C3D machine type AMD SEV Confidential Computing technology An operating system image that supports live migration

Looker - Feature: The Table Visualization Improvements preview feature is now available, and is disabled by default. When this preview feature is enabled, you can use the following features for table visualizations: Pagination Show/Hide Table Headers Table Borders Cell Highlighting Conditional formatting for string fields Note: When Cell Highlighting is enabled, conditional formatting is not available. Announcement: Starting with the Looker 26.8 release, which will release in May 2026, the following changes will occur: Looker (original) admins will no longer be able to access or manage the API credentials of their standard users Looker (original) admins will be able to manage which users have access to individually owned API credentials Users with access to API credentials will be able to access and manage their own API credentials. Before your Looker instance is upgraded to the Looker 26.8 release, your admin must follow the steps in the Migrating users to service accounts documentation page. This is to ensure admins can either create or migrate service accounts from existing standard users if they require access to users' API credentials. For more information, see the Discontinuing the admin capability to create, view, and manage API credentials for a standard user deprecation notice.

Cloud Logging - Feature: You can now ingest OTLP-formatted logs into Cloud Logging by using an OpenTelemetry Collector, an OTLP exporter, and the Telemetry API. For more information, see OTLP log ingestion overview. The Telemetry API for log ingestion is in Preview.

Bigtable - Feature: You can connect to Bigtable from Java applications and other reporting tools that support a generic JDBC adapter by using the Bigtable JDBC driver. This feature is generally available (GA). Feature: You can use protocol buffer (protobuf) schemas to query individual fields within protobuf messages stored as bytes in Bigtable. You can query your protobuf data using GoogleSQL for Bigtable, continuous materialized views, logical views, or BigQuery external tables. This feature is generally available (GA). Feature: You can use Gemini in Bigtable Studio to help you write GoogleSQL queries. This feature is available in Preview. For more information, see Write SQL with Gemini assistance.

Contact Center AI Platform - Full details on the release page.

Datastream - Feature: You can now use the Datastream remote MCP server to enable LLM agents to perform data-related tasks, such as managing and monitoring your streams, connection profiles, and stream objects. This feature is in Preview. Feature: Datastream is now integrated with Knowledge Catalog. This integration lets you search for and browse your Datastream resources, such as streams, connection profiles, and connectivity configurations directly in Knowledge Catalog. For more information, see the documentation.

Secret Manager - Feature: Parameter Manager supports the latest identifier, which lets you fetch the most recent parameter value without specifying a version ID. When you use the gcloud CLI or REST API, you can use latest to retrieve the most recent version of a parameter. For more information, see Access a parameter version.

GKE new features - Feature: Gateway API v1.5 is supported in GKE version 1.35.2-gke.1842000 and later. The GKE Gateway controller passes core conformance tests for this version of the Gateway API. Feature: GKE managed DRANET is now Generally Available (GA) for GKE version 1.35.2-gke.1842000 or later. GKE DRANET is a managed feature that implements the Kubernetes Dynamic Resource Allocation (DRA) API for high-performance networking. The GA release expands support beyond the preview phase to include the following hardware: NVIDIA GPU Instances: Support for instances starting from A3 Ultra, including A4, A4X, and A4X Max. Cloud TPU Instances: Support for TPU v6e and TPU v7x. For more information, see Allocate network resources by using GKE managed DRANET.

Cloud Trace - Feature: Google Cloud CLI lets you configure trace scopes, manage observability buckets, and set default observability settings. These features are in Public Preview. For more information, see the following documents: Configure trace scopes by using the Google Cloud console, the Google Cloud CLI, Terraform, or the Observability API. For more information, see Create and manage trace scopes. Manage trace storage by using the Google Cloud CLI or the Observability API. For more information, see Manage trace storage. Configure default settings by using the Google Cloud CLI, Terraform, or the Observability API. For more information, see Set defaults for observability buckets. Feature: Use Cloud Trace to troubleshoot your MCP server usage, tool failures, and latency causes. For more information, see Investigate MCP calls using Trace.

Backup and DR Service - Feature: Filestore enhanced backups are now generally available (GA). Enhanced backups are managed and stored by Backup and DR Service and provide immutable and indelible backups through backup vault, fine grained scheduling, backup protection against source instance deletion and centralized monitoring and reporting. For more information on protecting Filestore instances using Backup and DR Service, see Back up Filestore instances to a backup vault. Feature: Announcing the general availability (GA) of a 1-hour backup frequency for Compute Engine instances to support mission-critical workloads requiring a lower RPO. Backup and DR Service now supports Shielded VMs during the restore process, meaning no changes to org policies are required, maintaining a continuous security posture. You can now take application-consistent backups of disk data, enabling the execution of custom pre- and post-scripts. This allows databases and filesystems to be paused or flushed to disk properly before the backup, ensuring a consistent and usable state upon recovery. Note: This capability is available through Google Cloud CLI, Terraform, and API only. Security: Vulnerability Remediation: Addressed over 38 security vulnerabilities identified in internal and customer scans, including remediations for Kernel, OpenSSL, rsync, glibc, and OpenJDK packages. Fixed: HANA Backup and Restore Stability: Resolved HANA backup failures by improving the matching logic for alternate IP addresses and addressed restore job failures (error 5728) caused by missing configuration files during the database freeze phase. SAP IQ Log Management: Fixed an issue where SAP IQ log backups would fail due to invisible characters in temporary log file paths. Cleanup and Expiration Improvements: Resolved recurring cleanup job failures (error 1271) by ensuring failed backup images are assigned valid cloud credential IDs. Log Reliability and Diagnostics: Prevented root filesystem exhaustion by rate-limiting excessive UDSAgent logging during custom app activities. Additionally, the UDPINFO logging limit was increased to 10KB to prevent the truncation of large XML messages, facilitating better debugging. Feature: Expanded OS Support: Qualified Linux Change Block Tracking (CBT) support for multiple new RHEL 8 and 9 kernels, including versions 8.8, 8.10, 9.2, 9.4, and 9.7. Verbose Process Logging: Added enhanced verbose logging for fingerprint computation and verification to assist in troubleshooting verification failures.

Dataplex - Feature: You can now specify a custom execution identity for data quality and data profile scans. By default, scans are executed using the Service Agent. You can now use a custom service account (Bring Your Own Service Account) or End-User Credentials (EUC). Using a custom execution identity lets you enforce the principle of least privilege, use fine-grained BigQuery access controls, and unify scan processing costs directly under BigQuery. For more information, see Configure execution identity for data quality scans and Configure execution identity for data profile scans. Feature: The lightweight profiling mode for data profile scans is available in preview. The lightweight mode provides low-latency profile scans that return results in seconds, making it ideal for grounding AI agent responses and interactive data exploration. For more information, see Profiling modes. Announcement: Dataplex Universal Catalog is now called Knowledge Catalog. The API, client library, CLI, and Identity and Access Management (IAM) names remain unchanged.

Cloud Filestore - Feature: Filestore is integrated with Backup and Disaster Recovery (DR) Service allowing you to centrally manage your backups with advanced features for data protection. This feature is generally available for Filestore instances. For more information, see Backups overview.

Compute Engine - Feature: Generally available: Hyperdisk ML disks are supported by the following machine series: A3 Ultra C4D N4 N4D Hyperdisk ML offers the highest throughput of all Google Cloud Hyperdisk types, up to 2 TiB/s (2,097,152 MiB/s). For more information, see Hyperdisk ML overview.

Service Extensions - Feature: Google Kubernetes Engine (GKE) Gateway support for using extensions by using callouts to add custom logic into the load balancing processing path is in General Availability. For more information, see GKE extensions. Feature: When configuring extensions by using plugins or callouts, you can specify some request and connection attributes to forward to backend services. For more information, see supported attributes.

Load Balancing - Feature: Published service backends let you configure supported load balancers or regional Cloud Service Mesh to route traffic to published services through Private Service Connect endpoints. For more information, see Published service backends. This feature is in Preview.

Colab - Feature: SQL cells Generally available: You can use SQL cells to write, edit, and run SQL queries directly from your Colab Enterprise notebooks. For more information, see Use SQL cells.

Virtual Private Cloud - Feature: Private Service Connect consumers can configure supported load balancers or regional Cloud Service Mesh to access published services through Private Service Connect endpoints. This feature is available in Preview. For more information, see Published service backends.

Chronicle SOAR - Announcement: Release 6.3.82 is now available for all regions. Announcement: Release 6.3.83 is being rolled out to the first phase of regions as listed here. This release contains internal and customer bug fixes.