< ciso
brief />
Vendor and Hyperscaler Watch Banner

All news in category “Vendor and Hyperscaler Watch”

6816 articles · page 19 of 341

AWS Glue adds automatic Spark runtime upgrades

🔧 AWS Glue now offers automatic version upgrades for Apache Spark jobs, keeping runtimes current without manual coordination. The feature performs multi-level validations, seeks consent before running jobs on newer runtimes, and verifies upgraded runs against each job's established output profile. Setting a job's version to "Auto" enables hands-free upgrades, letting platform teams standardize runtimes and business units remain current while jobs needing code changes continue on their existing version.
read more →

AWS Backup adds EBS Snapshots Archive to LAG vaults

🔒 AWS Backup's logically air-gapped (LAG) vaults now support archiving Amazon EBS snapshots to Amazon EBS Snapshots Archive, a low-cost storage tier for rarely accessed snapshots. LAG vaults keep immutable, encrypted backups that are locked by default to meet long-term retention and compliance needs. You can enable this from the AWS Backup console, CLI, or SDKs by designating a LAG vault in a monthly (or less frequent) backup plan and enabling archive in the lifecycle settings. The feature is available in Regions where both services are offered.
read more →

AgentCore Runtime V2 Becomes Default in Many Regions

🆕 Starting today, new Amazon Bedrock AgentCore runtimes default to the next-generation AgentCore runtime, which restores sessions from initialized snapshots, scales compute and memory with workload, and scales idle agents to zero. Each session continues to run in hardware-isolated Firecracker microVMs and can handle heavy tasks such as compiling large codebases, driving browsers, or loading local models. Sessions now scale to 8 vCPU and 32 GB RAM, support container images up to 10 GB, and achieve much faster start times—under 2 seconds at P75 for a 2 GB image—while idle memory is reclaimed within minutes. Existing runtimes remain on V1 until moved; to use V2 explicitly set platformVersion to V2 when creating or updating a runtime, or set platformVersion to V1 to retain the current version. The new runtime is now the default across many AWS Regions including US, Canada, South America, Europe, and Asia Pacific.
read more →

Ornith 1.0 Models Added to SageMaker JumpStart

🚀 Ornith-1.0-9B and Ornith-1.0-35B are now available on Amazon SageMaker JumpStart, extending AWS’s collection of foundation models for agentic coding. The 9B dense model targets efficient single‑GPU deployment with a 256K token context window, while the 35B Mixture‑of‑Experts model balances performance and efficiency on a 24GB GPU. Both use reinforcement learning to optimize scaffold generation and solution rollouts for improved code generation.
read more →

DeepSeek V4 models added to SageMaker JumpStart

🚀 DeepSeek's DeepSeek-V4-Pro, DeepSeek-V4-Flash, and DeepSeek-V4.1-Flash are now available on Amazon SageMaker JumpStart as open-weight Mixture-of-Experts (MoE) language models under the MIT license. These models offer frontier performance across reasoning, coding, and agentic workloads while activating only a fraction of parameters per token for improved efficiency. Customers can deploy them from the SageMaker JumpStart catalog or via the SageMaker Python SDK to address long-context, multimodal, and cost-sensitive production use cases.
read more →

AWS Network Security Manager Expands to EU Sovereign Cloud

🔒 AWS announces general availability of AWS Network Security Manager in the AWS European Sovereign Cloud Region and in thirty-one additional regions worldwide. The service centralizes and automates network security policy deployment, supporting AWS WAF and AWS Shield in the sovereign region. It offers always-on automation to apply WAF protections to internet-facing assets, detects configuration drift, and enables centralized remediation to reduce operational overhead.
read more →

OpenSearch Serverless adds intelligent time series ingestion

🛠️ The next generation of Amazon OpenSearch Serverless now supports intelligent ingestion for time series collections, enabling direct ingestion of logs and events from Amazon S3 without operating an ingestion pipeline. Intelligent ingestion auto-detects data types, parses records, and indexes JSON, NDJSON, and JSON array logs via one-time ingestion jobs or continuous ingestion streams. Capacity scales with workload up to collection limits, and tiering keeps recent data hot while moving older data to a warm tier based on retention.
read more →

AWS Cloud WAN adds centralized VPC Flow Logs

🛰️ AWS Cloud WAN now supports VPC Flow Logs, enabling centralized visibility into IP traffic across global networks. Flow Logs capture per-flow details—IP addresses, ports, protocol, traffic volume, and packet-loss counters—plus Cloud WAN metadata like segment, Region, service insertion hops, and attachment details. Owners can enable logging for an entire core, specific edges, or individual attachments and publish records to CloudWatch Logs, S3, or Data Firehose. There is no additional Cloud WAN fee for Flow Logs; standard destination charges apply.
read more →

AWS simplifies getting started with secure defaults

🔒 The new AWS experience streamlines how you start building by applying sensible, secure defaults so projects are safe and manageable from day one. Features include Builder ID sign-in with MFA, an automatically managed AWS Organization and per-project boundaries, a single default Region, spend limits that pause projects at a ceiling, and preconfigured IAM roles to reduce initial setup. These defaults let developers focus on building while providing a secure foundation that can scale and be handed over to advanced controls when needed.
read more →

Amazon EKS increases etcd size for larger clusters

🔧 Amazon EKS now provides a larger 24 GB etcd database (up from 16 GB) for 4XL and 8XL Provisioned Control Plane clusters on EKS version 1.34 or later. The etcd store holds cluster state, and the 50% capacity increase allows more Kubernetes objects, supporting larger pods, jobs, events, and custom resources. Existing eligible 4XL and 8XL clusters receive the upgrade automatically, and the 24 GB database is available in all commercial AWS Regions where Provisioned Control Plane is offered.
read more →

Amazon Cognito adds user session management controls

🔒 Amazon Cognito now offers session management for user pools, enabling visibility and control over both SSO and token sessions created when users sign in. Administrators can list active sessions across devices and app clients, view details like device type and last activity, revoke individual or linked sessions, and perform global sign-out. App clients can have configurable SSO reuse durations and a maximum number of token sessions per user. The feature is available in Essentials and Plus plans across all Regions where Cognito is provided and manageable via Console, API, CLI, SDKs, and CloudFormation.
read more →

Cybersecurity Journeys: Advice to My Younger Self

🛡️ This final post in a four-part series collects the reflections and hard-won advice of 11 AWS security professionals about career growth, trust, and learning through discomfort. Conversations emphasize mentorship, delegation, and translating technical work for nontechnical audiences. Contributors share practical beginnings—from help desk to data center work—and urge early-career practitioners to ask questions and prepare for opportunity.
read more →

AWS Lambda MicroVMs Adds x86 Architecture Support

🖥️ AWS Lambda MicroVMs now supports the x86 CPU architecture, allowing you to run existing x86 applications, tools, and dependencies without recompilation. When creating a MicroVM image you can choose between x86 and Arm (AWS Graviton) to provide isolated MicroVMs per user or job. MicroVMs deliver VM-level isolation, fast launch/resume, and state preservation for use cases like AI coding assistants, CI/CD runners, and scanners. x86 support is available in most Regions where Lambda MicroVMs is offered, with a few regional exceptions.
read more →

EKS Auto Mode adds security groups for Pods

🔒 Amazon EKS Auto Mode now supports security groups for Pods, enabling assignment of dedicated Amazon EC2 security group rules to individual Pods. This allows different workloads on the same node to carry different security groups for network-level segmentation without separate node pools. You control assignments with a SecurityGroupPolicy resource; unmatched Pods retain node security groups. The feature is included with EKS Auto Mode at no extra cost and requires Kubernetes 1.36+ in supported Regions.
read more →

AWS Emulator now available for local development

🛠️ Starting today, AWS Emulator lets developers build and test AWS applications locally or in CI using the same AWS SDK and CLI calls. The emulator returns the same results and errors for supported operations, stores resource state in memory per-instance, and isolates ports, keys, and state for concurrent test runs. Services are generated from AWS API models and updated with AWS releases; unsupported operations return 501 Not Implemented.
read more →

Self‑hosted Gemma on TPUs for EV charging AI

🚗 Lemonflow builds real‑time AI agents for EV charging support and monitoring, self‑hosting the open Gemma 4 model on Google Cloud TPU v6e in the EU to achieve ultra‑low latency and deterministic control. They route live voice calls with priority, use custom vLLM optimizations and memory tiering to keep many agents warmed, and maintain automated fallback to GPUs and Gemini Enterprise. The platform converts diverse CPMS interfaces to a standard format, streams structured logs into the model, and executes remote charger actions while enforcing security controls and EU compliance.
read more →

AlloyDB AI adds in-database classify and extract

🧭 Google Cloud announces preview SQL functions ai.classify and ai.extract in the AlloyDB AI query engine, available via the google_ml_integration extension on PostgreSQL 18. These functions let you perform text classification and structured extraction directly in the database, supporting single- and multi-label classification, schema-driven JSON extraction, arrays and cursor batching, and few-shot prompting. Smart batching (AI Function Acceleration) groups inputs into micro-batches to improve latency and reduce token costs, enabling secure, operational workflows for healthcare, finance, legal, real estate, and knowledge graph use cases.
read more →

Anthropic Glasswing: Findings Overwhelming Maintainers

🔍 A VulnCheck report by Patrick Garrity critiques Anthropic’s Project Glasswing, finding a large volume of AI-flagged issues but few verified and fixed vulnerabilities. Anthropic’s dashboard lists 37,966 candidate findings with 5,276 validated unique vulnerabilities, yet only 217 have been fixed. Anthropic says triage capacity is a bottleneck and has launched an OSS Scanner to let maintainers self-triage, a move that has shifted significant workload to open source maintainers.
read more →

AWS Quick desktop and mobile apps reach Singapore

🖥️ The AWS Quick desktop app for macOS and Windows and the Quick mobile app for iOS and Android are now available in the Asia Pacific (Singapore) Region. The desktop app integrates with local files, calendar, email, and business apps, while the mobile app enables chat with Quick, activity feed review, and dashboard access. Conversations, context, and agents remain synchronized across web, desktop, and mobile so users can start agents on one device and continue on another. To get started, download the apps from the respective stores and consult the AWS Quick User Guide for more details.
read more →

Selecting QuePaxa for global consensus stability

🛡️ Cloudflare introduced Meerkat, a global control plane service, powered by a new consensus algorithm called QuePaxa. Traditional leader-based algorithms like Raft and Paxos perform poorly in high-latency, volatile wide-area networks because leader failures or network spikes stall progress. Meerkat needs a protocol that makes progress despite latency and partitioning, so Cloudflare chose QuePaxa for its leaderless resilience and improved stability in Internet-scale deployments.
read more →