Google Cloud Platform Newsletter

Check Archive for older issues

News

Announcing Ironwood TPUs General Availability and new Axion VMs to power the age of inference - Google Cloud’s compute portfolio now includes Ironwood TPUs and Axion-based N4A VMs and C4A bare metal.

More ways to build, scale, and govern AI agents with Vertex AI Agent Builder - Announcing new capabilities across the entire agent lifecycle to help developers build, scale, and govern AI agents using Vertex AI Agent Builder. Get started today.

The Data Engineering Agent is now in preview - The Data Engineering Agent in BigQuery, now in preview, automates many tedious development, maintenance, and troubleshooting tasks.

Upgrading Kubernetes versions just got safer with minor version rollback - In the event of an error when upgrading a Kubernetes cluster, a new minor version rollback feature lets you revert to a known-good state.

Unlock 2x better price-performance with Axion-based N4A VMs, now in preview - Lower the TCO of scale-out workloads like GKE and web servers. Google Cloud’s new N4A VMs deliver breakthrough price-performance and flexible shapes.

Announcing Axion C4A metal: Arm-based Axion instances for specialized use cases - C4A.metal is Google Cloud’s first Axion bare metal machine type, providing custom Arm CPUs for workloads like Android dev and automotive simulation.

Automating FinOps cost management policies using Workload Manager - Learn how to use Workload Manager to automate financial governance policies that help you incorporate FinOps practices.

Announcing the General Availability of Smarter, AI-powered Cost Anomaly Detection - Cost Anomaly Detection proactively identifies anomalies in near-real-time so you can avoid surprises, take swift action and control runaway costs.

AlloyDB accelerates AI with automated vector indexing and embedding - AlloyDB AI auto vector embeddings transform operational data into vector-search-ready data, and auto vector index configures optimized vector indexes.

A more native experience for Cloud TPUs with Ray on GKE - Ray on GKE has new features: label-based scheduling, atomic slice reservations, JaxTrainer, built-in TPU awareness (topologies/SPMD/metrics).

Google Cloud Europe establishes new European Advisory Board - A new European Advisory Board will help customers navigate complex regulatory landscapes, and foster a strong, sustainable digital economy.

Articles, Tutorials

Infrastructure, Networking, Security, Kubernetes

7 ways networking powers your AI workloads on Google Cloud - From API access to autonomous networking, learn more about how Google’s networking capabilities support your AI workloads.

Evolving Ray and Kubernetes together for the future of distributed AI and ML - Ray on Kubernetes now has new label-based scheduling, DRA for accelerators, writable cgroups, and vertical pod resizing for distributed AI/ML.

GTIG AI Threat Tracker: Advances in Threat Actor Usage of AI Tools - Google Threat Intelligence Group's findings on adversarial misuse of AI, including Gemini and other non-Google tools.

Preparing for Threats to Come: Cybersecurity Forecast 2026 - The Cybersecurity Forecast 2026 report contains forward-looking insights on AI, cybercrime, and nation-state activity.

Google Cloud Platform: Even FASTer Networking - The article discusses the evolution of Google Cloud's Fabric FAST, focusing on the new factory-based networking stage.

Google Cloud Deploy in Action - The article is a tutorial on using Google Cloud Deploy to automate application delivery to Google Kubernetes Engine environments.

Cloud CISO Perspectives: Recent advances in how threat actors use AI tools - GTIG has observed threat actors experimenting with deploying novel AI-enabled malware in active operations. Sandra Joyce details what you should know.

App Development, Serverless, Databases, DevOps

How Buildertrend Drives Innovation with Memorystore for Valkey - Buildertrend migrated from Memorystore for Redis to Memorystore for Valkey to access native cross-regional replication and Private Service Connect.

Inside Mercado Libre's multi-faceted Spanner architecture - Discover how Spanner became a core engine powering internal developer platforms, diverse data models, advanced analytics loops, and intelligent features for this large Latin American retailer.

Taming Python/FastAPI/Uvicorn Logs in GCP - The article discusses how to properly configure Python/FastAPI/Uvicorn logs in Google Cloud Logging to address issues such as incorrect log levels, fragmented multiline tracebacks, and misleading Uvicorn startup error messages.

Google Cloud Pub/Sub with the Rust SDK and Gemini CLI - The article details how to build a native Rust client for Google Cloud Pub/Sub using the Gemini CLI and Google Cloud Rust SDK.

Google Cloud Persistent Disks: A Comprehensive Tutorial - This tutorial provides a comprehensive guide on using Google Cloud Persistent Disks, covering creation, attachment to Compute Engine VMs, and formatting. It explains the importance of regions, zones, different disk types (HDD, SSD), and provides instructions for both Google Cloud Console and gcloud CLI.

Yes, we can have nice things! Using Gemini CLI in an Enterprise Environment - Speed vs. Safety Solved: How platform teams use the Gemini CLI Configuration Sandwich for enterprise-ready AI tooling.

Firestore Extension for the Gemini CLI - Query, manage, and visualize your data directly from your terminal using AI.

Big Data, Analytics, ML&AI

How Ericsson achieves data integrity and superior governance with Dataplex - Ericsson moved from foundational practices to a sophisticated, business-enabling data governance framework using Google Cloud’s Dataplex Universal Catalog, turning data from a simple resource into a strategic asset.

BigQuery Join Order: Let the Optimizer Win - Subtitle: How to write intent-driven SQL BigQuery can optimize — plus when to step in with hints and materialization.

Migrating Apache Iceberg Tables to GCS and BigLake Iceberg REST Catalog - The article outlines the process of migrating Apache Iceberg tables to Google Cloud, leveraging BigLake Metastore for unified data management.

Easy AI workflow automation: Deploy n8n on Cloud Run - With just a few commands, you can deploy n8n to Cloud Run and have it up and running, ready to supercharge your business with AI workflows that can manage spreadsheets, read and draft emails, and more.

Boosting LLM Performance with Tiered KV Cache on Google Kubernetes Engine - Boost LLM inference performance with LMCache on Google Kubernetes Engine. Discover how tiered KV cache expands NVIDIA GPU HBM with CPU RAM and local SSDs, significantly improving context length, concurrency, and throughput for large language models.

From silicon to softmax: Inside the Ironwood AI stack - Learn about the core components of Google's AI software stack woven into the Ironwood TPU, including JAX and PyTorch ecosystems, the XLA compiler.

How to use GCP Vertex AI Vizier Hyperparameter Optimization - 50 GCP instances managed by Vertex AI Vizier; Google’s “black-box” Bayesian hyperparameter optimization service (HPO).

Building Collaborative AI: A Developer's Guide to Multi-Agent Systems with ADK - Unlock the power of collaborative AI with Google's Agent Development Kit (ADK). This guide explores Multi-Agent Systems (MAS), ADK's agent types (LLM, Workflow, Custom), hierarchical structures, and communication mechanisms for building robust, scalable, and intelligent AI solutions.

Quick Guide to ADK Callbacks - The article provides a guide to Agent Developer Kit (ADK) callbacks, which are checkpoints in an agent's lifecycle that allow developers to hook into key stages like receiving input, invoking models/tools, and returning output.

Your First AI Application is Easier Than You Think - Learn how to build your first AI-powered application with Google's Gemini model. This codelab guides developers with no prior AI experience through creating an interactive travel chatbot, covering connection, system instructions, and real-world data integration.

Building AI Agents with the GO Agent Development Kit (ADK) - Native GO Agent Coding with the GO ADK.

Go SDK Power-Up: ADK Support Arrives in MCP Toolbox - Easily build smart agents based on your enterprise data.

Google Just Made RAG Ridiculously Easy with the New File Search Tool - Google has introduced a new File Search Tool within the Gemini API that simplifies RAG (Retrieval-Augmented Generation) implementations. The tool allows users to upload various file formats, and Gemini then intelligently examines these files to answer questions, eliminating the need for separate vector databases or retrieval pipelines.

ADK architecture: When to use sub-agents versus agents as tools - As you build sophisticated multi-agent AI systems with the Agent Development Kit (ADK), a key architectural decision involves choosing between a sub-agent and an agent as a tool. This choice fundamentally impacts your system's design, how well it scales, and its efficiency.

Various

Build software sustainably in the AI era - How AI and cloud innovation can help organisations build more efficient, resilient, and sustainable software systems.

Slides, Videos, Audio

Agent Factory Recap: Build AI Apps in Minutes with Google's Logan Kilpatrick - Learn how to build AI apps in minutes with Google AI Studio and Logan Kilpatrick from Google DeepMind. Discover vibe coding, grounding with Google Maps, and the future of AI development in this Agent Factory recap.

Security Podcast - #250 The End of "Collect Everything"? Moving from Centralization to Data Access?

GCP Bytes Podcast - #29 In this episode we discuss; AWS Postmortem, MS Outage, Formae, GDG, Q3 Results, Anthropic Deal, DC Spend, AWS Layoffs, ACCC & Microsoft, Optus Inquiry, Optus & Nokia, 2tb SQL Exposed, A4X Instances, Claude Skills Hack, Veo3.1.

Releases

Apigee API Hub - Filter APIs by user-defined attributes You can now filter APIs using your custom, user-defined attributes from the APIs page in the Google Cloud console. Issue API hub provisioning fails in data residency enabled Apigee organizations Currently, API hub can't be provisioned within an Apigee organization that has data residency enabled.

BigQuery - The research paper ARIMA_PLUS: Large-scale, Accurate, Automatic and Interpretable In-Database Time Series Forecasting and Anomaly Detection in Google BigQuery is now publicly available. You can use the MATCH_RECOGNIZE clause in your SQL queries to filter and aggregate matches across rows in a table. Announcement The BigQuery Data Transfer Service for Google Ads now supports Google Ads API v21. You can now generate data insights when you create a DataScan using the Dataplex API. You can now generate table and column descriptions in all supported Gemini languages when you generate data insights. You can now use custom organization policies with the BigQuery migration service to allow or deny specific operations during a BigQuery migration to meet your organization's compliance and security requirements.

Bigtable - You can use protocol buffer (protobuf) schemas to query individual fields within protobuf messages stored as bytes in Bigtable.

Chronicle Security Operations - Change MITRE ATT&CK coverage dashboard is now available The new MITRE ATT&CK coverage dashboard lets you measure your security posture against the MITRE ATT&CK framework, helping you: Assess threat coverage Identify gaps Prioritize security efforts.

Chronicle SOAR - Announcement Release 6.3.67 is being rolled out to the first phase of regions as listed here. Announcement Release 6.3.66 is now available for all regions.

Cloud Composer - We strongly recommend to use highly resilient environments for production use cases. A new Cloud Composer release has started on November 05, 2025. Security Fixed the CVE-2025-49844 vulnerability. Change New Airflow builds are available in Cloud Composer 3: composer-3-airflow-2.10.5-build.19 (default) composer-3-airflow-2.9.3-build.39. Change New images are available in Cloud Composer 2: composer-2.15.3-airflow-2.10.5 (default) composer-2.15.3-airflow-2.9.3.

Compute Engine - Generally available: N4D VMs are powered by the fifth generation AMD EPYC Turin processor and Titanium I/O offload processing. Preview: N4A VMs, powered by Google's next generation Axion processor built on the Arm Neoverse N3 platform are in Preview. Generally available: You can verify which reservation a VM is consuming and view a list of VMs consuming a reservation. Public preview: You can configure a regional managed instance group (MIG) to allow a VM repair in an alternate zone when the MIG can't repair the VM in its original zone.

Contact Center AI Platform - Announcement Google Cloud CCaaS 3.41 We've released version 3.41 of Google Cloud CCaaS. Nested disposition lists. You can now organize your disposition lists by grouping them into nested folders, making it easier for agents to find the disposition codes they need. Skip language selection and IVR menu readout You can configure your call flow to skip language selection and the IVR menu readout. Transfer an email session to a queue Agents can now transfer an email session to a queue. Virtual agent to virtual agent direct chat transfers You can configure Dialogflow payloads to transfer chat sessions from one virtual agent directly to another virtual agent, using the destination virtual agent's agent ID. Fixed We addressed the following issues in this release: Fixed an issue where hyperlinks in the after-hours message of the web SDK were broken.

Dataproc Serverless - Announcement Serverless for Apache Spark: Apache Spark upgrade to version 3.5.3 for the 1.2 and 2.2 Serverless for Apache Spark runtime versions announced on October 13, 2025 has been rolled back.

Document AI - Gemini layout parser is in Preview. v1 & v1beta3. Layout parser support for DOCX, PPTX, XLSX and XLSM file types in Document AI is in General Availability (GA).

Cloud Healthcare API - VersionedStorageSizeBytes was added to the output of GetFhirStoreMetrics.

Google Kubernetes Engine - Change (2025-R46) Version updates GKE cluster versions have been updated. Security (2025-R46) Security updates This release includes new GKE versions that use updated Container-Optimized OS images. In GKE version 1.34.1-gke.2037001 and later, the GKE logging agent in your clusters can process logs up to two times faster per node than in version 1.33 and earlier. In version 1.34.1-gke.1829001 and later, GKE can auto-create multiple node pools concurrently to improve the speed with which multiple new node pools become ready. In GKE version 1.35 and later, GKE rejects anonymous requests to cluster endpoints (except for the livez, /healthz, and /readyz health check endpoints) by default for all new Autopilot or Standard clusters.

Load Balancing - GRPC_WITH_TLS health checks are used for health checking gRPC backends with TLS enabled.

Looker - Looker (Google Cloud core) and Looker (original) changes. Other Looker 25.20 is expected to include the following changes, features, and fixes: Expected Looker (original) deployment start: Thursday, November 6, 2025 Expected Looker (original) final deployment and download available: Thursday, November 13, 2025 Expected Looker (Google Cloud core) deployment start: Thursday, November 6, 2025 Expected Looker (Google Cloud core) final deployment: Tuesday, November 18, 2025. Looker (Google Cloud core) and Looker (original) changes. Breaking Looker now recognizes results from the YEAR function as a number and not a date in MySQL 8.4. Looker (Google Cloud core) and Looker (original) changes. Breaking The System Activity fields sql_query.sql and sql_text.text have been limited to a length of 1.9 MB. Looker (Google Cloud core) and Looker (original) changes. The Max Number of PDT Builder Connections limit has been increased from 10 to 100. Looker (Google Cloud core) and Looker (original) changes. Looker 25.20 contains the following accessibility improvements: The character count for the custom message field of a schedule is now accessible to screen readers. Looker (Google Cloud core) and Looker (original) changes. Fixed An issue has been fixed where a measure of type: period_over_period could be referenced in the sql parameter for a measure of type: number. Looker (Google Cloud core) and Looker (original) changes. Fixed An issue has been fixed where the Clear field keep filter button on an Explore could incorrectly become disabled when a user used the field search bar in the field picker. Looker (Google Cloud core) and Looker (original) changes. Fixed An issue has been fixed where SQL Runner could return a NoMethodError when trying to query a LookML model that doesn't exist in the user's development mode. Looker (original) changes. Fixed An issue has been fixed where disabling the Favoriting LookML Dashboards Labs feature prevented users from moving LookML dashboards in folders. Looker (original) changes. A new Labs feature, Internal Dashboard Theming, lets users apply predefined themes to a Looker dashboard. Looker (original) changes. The Visual Drilling feature is now out of Labs and generally available on the Admin > General Settings page. Looker (Google Cloud core) changes. The Visual Drilling feature is now generally available on the Admin > General Settings page.

Media CDN - Media CDN supports flexible shielding.

Cloud Monitoring - You can now view the topology for applications that you register with App Hub.

NetApp - Google Cloud NetApp Volumes supports the block storage feature with iSCSI protocol in allow-listed General Availability (GA) for the Flex service level of the Unified type.

Cloud Run - Use dual-stack subnets with IPv6 to let your Cloud Run resources send IPv4 and internal IPv6 traffic to a VPC network with Direct VPC egress, and send external IPv6 traffic to the public internet.

Security Command Center - You can use customer-managed encryption keys (CMEK) organization policies with Security Command Center. Security Command Center Risk Engine supports Cloud Run attack paths for the following high-value resources: run.googleapis.com/Job run.googleapis.com/Service. Compliance Manager supports the Security Command Center Premium tier at the organization level.

Sensitive Data Protection - The OBJECT_TYPE/PERSON/PASSPORT and OBJECT_TYPE/PERSON/PHOTO_ID_CARD infoType detectors are available in global and the asia, europe, and us multi-regions.

Service Extensions - Cloud Load Balancing callouts have full duplex streaming support.

Cloud SQL MySQL - Cloud SQL now supports the automatic minor version upgrade for Cloud SQL for MySQL 8.0.35 or later instances.

Cloud SQL Postgres - The rollout of the following extension versions, plugin versions, and extension support is underway: Extensions and plugins plpgsql_check is upgraded from 2.8.1 to 2.8.3 for PostgreSQL versions 14 and later.

Cloud Trace - You can now collect, view, and analyze prompts and responses from your agentic applications when they are built with the Agent Development Kit (ADK).

Virtual Private Cloud - VPC Network Peering supports peering connections in consensus mode.