All news in category “Vendor and Hyperscaler Watch”

🔔 Amazon Connect Customer now offers points-based scoring to evaluate human and AI agent performance, giving managers greater flexibility to assign importance to individual criteria. With this model, scores are the sum of points earned rather than percentages that must total 100. Managers can combine criteria, exclude items from scoring, or award bonus points to reflect business priorities. The feature is available in all regions where Amazon Connect Customer is offered.

🔔 AWS Glue 5.1 is now available in the AWS European Sovereign Cloud Region, bringing core engine upgrades to Apache Spark 3.5.6, Python 3.11, and Scala 2.12.18 for improved performance and security. The release updates open table format support for Apache Hudi 1.0.2, Apache Iceberg 1.10.0, and Delta Lake 3.3.2, and adds Iceberg format 3.0 features such as default column values and row lineage tracking. Lake Formation now enforces fine-grained access control for write DML and DDL operations in Spark DataFrames and Spark SQL, and full-table access control is added for Hudi and Delta Lake tables.

🛠️ Google Cloud announces Fault Injection Testing (FIT) in public preview to help teams automate failure testing and validate application resilience. FIT lets you create experiment templates to inject targeted faults such as Cloud SQL failovers and degraded Layer 7 traffic, with an automated dry run to verify affected resources and permissions. Experiments run for a defined duration with stop-and-revert controls, and Google recommends using FIT in non-production during preview. Access is via the Cloud console, gcloud CLI, or REST API; request preview through your account team.

🚦Uber adopted application awareness on Cloud Interconnect to prioritize business-critical traffic across its hybrid networks. As an early design partner, Uber deployed the feature in multiple locations to classify and queue application traffic using DSCP marking, protecting low-latency services during congestion. The approach improved bandwidth utilization, reduced the need for costly overprovisioning, and enabled safer migration of strategic workloads to Google Cloud.

🔍 This post explains how Microsoft Foundry helps reduce the cost per successful AI outcome by optimizing each model request at runtime. It outlines four levers—model routing and deployment choices, caching, prompt and agent optimization, and observability—that teams can apply, measure, and reverse if necessary. The guidance emphasizes matching model and deployment choices to task complexity, using caching and fine-tuning to lower repeated costs, and relying on observability and evaluation to validate savings without degrading quality.

📘 This article explains how to publish and govern Open Knowledge Format (OKF) bundles across an organization by mapping OKF concepts onto Google Cloud's Knowledge Catalog. It outlines a one-time setup and a single push workflow using sample code and the kcmd CLI to register EntryGroups, EntryTypes, and an okf AspectType carrying OKF v0.2 signals. The approach makes bundles discoverable, searchable, and governed by existing IAM policies alongside BigQuery, Cloud Storage, and other cataloged resources.

🚀 This post outlines practical strategies for dynamic capacity management to support large-scale AI and agent workloads, emphasizing predictable cost and performance. It describes three implementations: scheduling capacity for planned events, creating automated fallback plans with managed instance groups, and automating the lifecycle with GKE and Custom ComputeClasses. The guidance highlights tools like Dynamic Workload Scheduler, instance flexibility, Spot VMs, Hyperdisk, and dynamic resource allocation to improve utilization and resilience.

🧾 Google Cloud announces expanded billing flexibility and cost controls for agent workloads across Gemini Enterprise and developer tools like Google Antigravity and Android Studio. New options include pay-as-you-go consumption, pooled quotas, Flexible Savings Plans with 10–20% token discounts, and deferred execution pricing for off-peak discounts. Admins gain consolidated spend guardrails, hard monthly caps, anomaly detection, and centralized billing visibility to align FinOps with agent-driven innovation.

🛡️ Check Point introduces contextual AI protection that evaluates an agent’s full activity—intent, encountered information, prior actions and applicable policy—to prevent harmful or unauthorized outcomes before they execute. Traditional controls detect isolated risks like prompt injection or data exposure, but contextual protection links multi-step behavior to identify dangerous outcomes that no single rule would catch. Running in production, it enforces decisions in about 50 ms to block data leaks, unauthorized uploads, destructive commands and improper permission changes without slowing agents.

🔒 Microsoft is testing new privacy controls in Windows 11 Insider Experimental Preview Build 26340.9233 that let users manage camera, microphone, and location access per desktop app. Previously, these permissions were controlled by device-wide settings. Insiders can adjust access under Settings > Privacy & security using dedicated toggles. Microsoft cautions users to grant access only to trusted apps, noting some apps may appear unsigned or under different names.

🛡️ Microsoft warns that the interval between vulnerability disclosure and exploitation is rapidly shrinking, outpacing many organizations' ability to safely deploy patches. Igor Sakhnov, Azure Networking GM, urges a shift to network-level controls as a temporary control plane to limit exposure while patches are tested and rolled out. Analysts note the model suits mature cloud environments but faces practical challenges such as poor asset visibility and risks that temporary measures become permanent liabilities.

🔒 AWS Backup now supports cross-Region copies and logically air-gapped vaults for Amazon DocumentDB in nine additional Regions including Asia Pacific, Europe, Africa, and Israel. You can copy backups using backup plans or on-demand copy jobs to meet disaster recovery, continuity, and compliance needs. Logically air-gapped vaults provide immutable, isolated backups that are locked by default and encrypted with AWS-owned or customer-managed keys. Share vaults for recovery via AWS Resource Access Manager and protect access with multi-party approval.

🛡️ This post announces an independent assessment showing how Landing Zone Accelerator on AWS (LZA) can deploy multi-account environments with coverage for Australian Government ISM controls. The analysis, performed by AWS Partner gwi.digital, evaluates LZA Universal Configuration against 1,081 ISM controls and introduces the Controls Acceptance Testing Suite (CATS) for automated validation. Results highlight that LZA addresses 256 addressable controls, achieving Full or Partial coverage for 234, and offers mappings and tooling to accelerate IRAP assessment readiness for Australian customers.

🚀 Amazon EC2 C8id and M8id instances with up to 22.8 TB of local NVMe SSD storage are now available in additional AWS Regions. C8id instances launched in Asia Pacific (Sydney) and Canada (Central); M8id instances launched in Asia Pacific (Mumbai) and Canada (Central). These instances provide significantly higher vCPU, memory, storage, and I/O performance versus prior sixth-generation instances and support EFA on larger sizes.

🚀 Amazon EC2 R8id instances are now available in additional regions, including Asia Pacific (Mumbai, Malaysia, Sydney), Canada (Central), and Europe (Ireland, Stockholm). These instances provide up to 22.8TB of NVMe SSD local storage and 3.3x higher memory bandwidth, delivering up to 43% better performance versus R6id for memory-intensive workloads. Customers can adjust network and EBS bandwidth by 25% using EC2 instance bandwidth weighting and can use EFA on larger sizes. R8id instances are purchasable via Savings Plans, On-Demand, and Spot.

📡 AWS IoT Core now includes an InfluxDB rule action that sends time-series data from connected IoT devices directly to InfluxDB-compatible databases. The integration converts device telemetry to InfluxDB line protocol and supports writing to Amazon Timestream or self-hosted InfluxDB. It offers both device-side and server-side batching to optimize cost and throughput. The feature is available in all AWS Regions where Amazon Timestream for InfluxDB is supported.

🚀 AWS Batch now supports Amazon ECS Managed Instances (ECS MI) as a compute option, enabling GPU-accelerated and compute-intensive batch workloads on AWS-managed infrastructure. With AWS handling AMI updates, security patching, and instance lifecycle, customers can avoid operating their own EC2 fleets. Create a compute environment via the AWS Batch CreateComputeEnvironment API or the management console, specify instance types and networking in managedInstancesProvider, and submit jobs using On-Demand, Spot, or reserved capacity. This capability is available in all Regions where AWS Batch is offered; refer to the AWS Batch User Guide for details.

🟢 Starting today, you can add Amazon EC2 Capacity Blocks for ML and interruptible Capacity Reservations to Capacity Reservation Resource Groups. Previously limited to On-Demand Capacity Reservations, resource groups now accept any reservation type, simplifying instance launches across reserved capacity. Use a Capacity Reservation Resource Group in launch requests, and when using EC2 Fleet or Auto Scaling you can set prioritization and automatic fall back to On-Demand. This feature is available in supported AWS Regions with no additional charges.

🔒 AWS Lambda MicroVMs now support AWS PrivateLink, enabling private connectivity from Amazon VPC resources to Lambda MicroVMs without exposing traffic to the public internet. This helps regulated workloads in financial services, healthcare, and government meet strict network isolation requirements when building with MicroVMs. PrivateLink VPC Endpoints allow creation and invocation of MicroVM APIs and access to each MicroVM’s HTTP endpoint. Endpoints can be created via the Console, CLI, CloudFormation, or SDKs and are supported in all Regions where Lambda MicroVMs are available.

🔔 Amazon RDS for PostgreSQL now supports minor releases 18.6, 17.11, 16.15, 15.19, and 14.24. We recommend upgrading to these versions to address prior CVEs and to benefit from community bug fixes and improvements. You can apply upgrades during scheduled maintenance with automatic minor version upgrades, and orchestrate phased rollouts using AWS Organizations Upgrade Rollout Policy. Use Blue/Green deployments to minimize downtime for upgrades.