All news in category “Vendor and Hyperscaler Watch”

🔒 AWS Site-to-Site VPN now supports configurable tunnel bandwidth up to 5 Gbps, a 4x increase over the previous 1.25 Gbps limit. The update reduces the need to deploy complex protocols such as ECMP to aggregate tunnels, simplifying high-throughput hybrid connectivity for migrations, analytics, and disaster recovery. The capability is available in most commercial and GovCloud (US) Regions with a few regional exceptions.

📊 Amazon CloudWatch metrics are now available for S3 Tables, providing visibility into storage, maintenance, and request activity. Metrics include daily storage and object counts, compaction bytes/objects processed, and minute‑level request measurements for operations, data transfer, errors, and latency. You can access these metrics via the CloudWatch console, AWS CLI, or CloudWatch API at the bucket, namespace, and individual table level; they are available in all Regions where S3 Tables is offered.

🚀 Cloudflare has made remote bindings generally available, letting local Workers connect to live resources such as R2 buckets, D1 and KV namespaces without deploying. Developers can enable a binding with "remote: true" in Wrangler v4.37.0 and use existing Wrangler OAuth credentials to access production data. The local workerd runtime proxies JS API calls to remote service bindings (including JSRPC via Cap’n Web websockets), and tooling like the Vite plugin and vitest-pool-workers can use utilities such as startRemoteProxySession to join remote sessions.

🔒 Google on Tuesday introduced Private AI Compute, a cloud privacy capability that aims to deliver on-device-level assurances while harnessing the scale of Gemini models. The service uses Trillium TPUs and Titanium Intelligence Enclaves (TIE) and relies on an AMD-based Trusted Execution Environment to encrypt and isolate memory on trusted nodes. Workloads are mutually attested, cryptographically validated, and ephemeral so inputs and inferences are discarded after each session, with Google stating data remains private to the user — 'not even Google.' An external assessment by NCC Group flagged a low-risk timing side channel in the IP-blinding relay and three attestation implementation issues that Google is mitigating.

🚀 Starting today, Amazon EC2 F2 instances — the second-generation FPGA-powered instances featuring an FPGA with 16 GB of high-bandwidth memory (HBM) — are available in four additional regions: Europe (Frankfurt), Asia Pacific (Tokyo and Seoul), and Canada (Central). F2 delivers substantial hardware upgrades over F1, including up to 192 vCPUs, 2 TB system memory, 7.6 TiB SSD, and 100 Gbps networking. These instances target genomics, multimedia processing, big data, and network acceleration workloads and can be purchased On-Demand or via Savings Plans.

📈 The Amazon Managed Service for Prometheus collector now supports discovery and scraping of Prometheus metrics from Amazon Managed Streaming for Apache Kafka (MSK) clusters without deploying agents. The agentless collector can target metrics exposed via the JMX exporter and the Node exporter, covering host-level, JVM-level, and broker-specific telemetry. This simplifies open monitoring for MSK, improves availability and scalability, and is available in all commercial regions where the service is offered.

💬 The AWS Builder Center introduces Spaces, a community collaboration feature that lets builders create and join topic-focused groups to share knowledge and collaborate on AWS solutions. Spaces supports three visibility modes — Public, Private, and Invite-Only — with membership controls, approval workflows, and invite capabilities. Members can post text and images, comment, react, and search discussions, while owners and admins self-moderate content. The feature includes moderation tools and multi-language support across 16 languages to keep conversations focused and accessible.

🔍 AWS has extended the Cost and Usage Report (CUR 2.0) to surface hourly, resource-level billing information for capacity reservations including EC2 On-Demand Capacity Reservation (ODCR) and EC2 Capacity Blocks for ML. CUR 2.0 now tags capacity-related line items as Reserved, Used, or Unused, enabling precise coverage and utilization calculations. The enhancement helps identify idle reservations and attribute reservation costs to resource owners for cost optimization.

⚠️ Microsoft confirmed that Windows 11, version 23H2 Home and Pro editions reached end of servicing on November 11, 2025; the November 2025 monthly security update is the last patch for those SKUs. Devices running those editions will no longer receive monthly security or preview updates protecting against the latest threats. Users are advised to upgrade to Windows 11, version 25H2, available to eligible devices via Settings > Windows Update.

🛠️ AWS Parallel Computing Service (PCS) now supports Slurm CLI Filter plugins, letting administrators extend and modify how Slurm evaluates and schedules HPC jobs without changing Slurm source code. With CLI Filter plugins, you can enforce custom submission policies — validate required flags, reject submissions missing attributes, or adjust job parameters at submission. This capability is available in all Regions where PCS is offered.

🚀 Lightricks rewrote its training stack in JAX to scale high-performance video diffusion models on TPUs after hitting limits with PyTorch/XLA. The migration enabled reliable sharding, fixed FlashAttention and data-loading issues, and delivered linear scaling across small and large TPU pods. These improvements translated to ~40% more training steps per day, faster iteration, and doubled team productivity. Their stack leverages Flax, Optax, Orbax, and the MaxText blueprint for robust, testable, and efficient large-scale training.

🔍 In early 2024 Google introduced native vector search in BigQuery, embedding semantic search directly into the data warehouse to remove the need for separate vector databases. Users can create indexes with a simple CREATE VECTOR INDEX statement and run semantic queries via the VECTOR_SEARCH function or through Python integrations like LangChain. BigQuery provides serverless scaling, asynchronous index refreshes, model rebuilds with no downtime, partitioned indexes, and ScaNN-based TreeAH for improved price/performance, while retaining row- and column-level security and a pay-as-you-go pricing model.

🏆Fortinet’s FortiGate Rugged series (FGR-50G-5G and FGR-70G-5G) earned the Red Dot Product Design Award for its fanless industrial design, integrated 5G, and purpose-built ASIC performance. Engineered for OT and critical infrastructure, the appliances combine thermal resilience, shock and moisture protection, and low-latency security functions including next-generation firewalling, SD-WAN, VPN, and AI-driven threat detection. The recognition underscores Fortinet’s focus on precision engineering and durable, field-ready security.

🚀 Amazon EC2 C8gd instances are now available in Europe (London) and Canada (Central), while M8gd and R8gd sizes have expanded to South America (Sao Paulo) and Europe (London), respectively. Powered by AWS Graviton4, these instances deliver up to 30% better performance versus Graviton3 and offer up to 11.4 TB NVMe local storage and EFA on select sizes. Customers can also adjust network and EBS bandwidth by 25% via instance bandwidth weighting.

⚙️ Amazon EC2 High Memory U7i-6tb instances are now available in Europe (Stockholm and Ireland). The u7i-6tb provides 6TB of DDR5 memory and 448 vCPUs, with up to 100 Gbps for EBS and network bandwidth and support for ENA Express. Powered by custom 4th-gen Intel Xeon (Sapphire Rapids), these instances target mission‑critical in‑memory databases such as SAP HANA, Oracle, and SQL Server.

🔧 Mountpoint for Amazon S3 is now included in Amazon Linux 2023, making it straightforward to install, update, and mount S3 buckets with a single command. Previously, users downloaded the Mountpoint package from GitHub, resolved dependencies, and managed updates manually; inclusion in AL2023 streamlines that workflow. The open source project is backed by AWS and offers 24/7 AWS cloud support for Business and Enterprise Support customers—consult the repository and documentation to get started.

🚀 Amazon Web Services has made EC2 C6id instances available in Europe (Milan) and R6id instances available in Africa (Cape Town). Powered by 3rd-generation Intel Xeon Scalable Ice Lake processors (3.5 GHz all-core turbo) and up to 7.6 TB of local NVMe SSD, these Nitro-based instances deliver high compute, memory access, and low-latency storage. Use cases include media processing, distributed in-memory caches, in-memory databases, data logging, and real-time analytics. Customers can purchase capacity via Savings Plans, Reserved, On-Demand, and Spot, and provision using the AWS CLI and SDKs.

🔔 Amazon CloudWatch now lets teams create threshold-based composite alarms that trigger only when a specified subset of monitored resources meet a condition. Using the new AT_LEAST function, you can define fixed counts or percentages — for example, at least two of four volumes low on capacity or 50% of hosts with high CPU — to reduce alert noise. The capability is available in all commercial AWS regions, AWS GovCloud (US), and China Regions; composite alarms pricing applies.

🚀 Google details a broad set of GKE and Kubernetes enhancements announced at KubeCon to address agentic AI, large-scale training, and latency-sensitive inference. GKE introduces Agent Sandbox (gVisor-based) for isolated agent execution and a managed GKE Agent Sandbox with snapshots and optimized compute. The platform also delivers faster autoscaling through Autopilot compute classes, Buffers API, and container image streaming, while inference is accelerated by GKE Inference Gateway, Pod Snapshots, and Inference Quickstart.

🛡️ Agent Sandbox is a new Kubernetes primitive designed to run AI agents with strong, kernel-level isolation. Built on gVisor with optional Kata Containers and developed in the Kubernetes community as a CNCF project, it reduces risks from agent-executed code. On GKE, managed gVisor, container-optimized compute and pre-warmed sandbox pools deliver sub-second startup latency and up to 90% cold-start improvement. A Python SDK and a simple API abstract YAML so AI engineers can manage sandbox lifecycles without deep infrastructure expertise; Agent Sandbox is open source and deployable on GKE today.