< ciso
brief />
Vendor and Hyperscaler Watch Banner

All news in category “Vendor and Hyperscaler Watch”

6816 articles · page 22 of 341

AWS Glue adds 100 native connectors for SaaS data

🔌 AWS Glue now provides 100 new native connectors to simplify access to data in popular SaaS and REST API-based applications. Previously requiring custom code, teams can now create a single connection in the AWS Glue Data Catalog and reuse it across ETL jobs while preserving existing permission models. The connectors cover domains such as generative AI, finance, marketing, HR, product analytics, engineering, and collaboration, enabling data to be loaded into Amazon S3 or Amazon Redshift for analytics. These connectors are available in all AWS Commercial Regions on Glue versions 5.0 and above.
read more →

AWS STS Enables Cross‑Partition IAM Federation

🛡️ AWS Security Token Service now enables workloads to use their IAM identity in one partition to obtain temporary credentials for a role in another partition, eliminating the need for long‑term keys. Partitions, such as the commercial and GovCloud regions, have separate IAM instances that previously required long‑term credentials for cross‑partition access. The feature uses existing GetWebIdentityToken and AssumeRoleWithWebIdentity APIs, supports governance and CloudTrail auditing across partitions, and is available at no extra charge across commercial, GovCloud, China, and Germany sovereign regions.
read more →

Defending AgentCore Memory with Multi‑Layer Validation

🛡️ This post describes a composable, pre-write validation pipeline for Amazon Bedrock AgentCore Memory that augments existing controls like IAM, Cedar policies, and AgentCore guardrails. The solution attaches as a Lambda interceptor on AgentCore Gateway and scores writes using four signals — including a primary semantic-drift signal powered by Amazon Titan Text Embeddings V2 — to decide ALLOW, QUARANTINE, or DENY. It includes an AWS CDK deployment, Cedar policy examples, quarantine handling, alerting via SNS, test results from 11 scenarios, and latency/cost guidance for production use.
read more →

AWS rebrands Amazon Quick as AWS Quick

🔒 Amazon Quick is now AWS Quick, reflecting its integration as an AWS service with built-in enterprise security and governance. The new name appears across the Quick website, apps, and documentation, while existing customers do not need to take any action. Amazon Quick Sight is similarly renamed AWS Quick Sight, with no change to dashboards, datasets, or pricing. Quick remains a workplace AI platform with agent, app, and automation capabilities that maintain data residency and compliance.
read more →

Anthropic Claude Mythos 5.2 Now in AWS GovCloud

🔒 Amazon Bedrock now provides Claude Mythos 5.2 to eligible customers in AWS GovCloud (US). The model supports cybersecurity, software coding, and complex reasoning tasks, identifying sophisticated vulnerabilities and assessing exploitability. Access is restricted to allow-listed organizations via Anthropic's Cyber Verification Program. This release brings Anthropic's most advanced cybersecurity-capable model to GovCloud.
read more →

Amazon Connect adds evaluation form templates

📝 Amazon Connect Customer now offers prebuilt evaluation form templates to speed deployment of automated reviews for human and AI agents. Templates cover common scenarios—customer service, inbound sales, and self-service—and assess behaviors and business outcomes such as customer satisfaction and attrition risk. Managers can use templates as-is or customize them before deploying. The feature is available in multiple AWS Regions.
read more →

Amazon MWAA Adds Native DuckDB Operator Support

🐘 Amazon Managed Workflows for Apache Airflow (Amazon MWAA) now includes native DuckDB operators in Provisioned environments running Apache Airflow 3.3.1. DuckDB is preinstalled and preconfigured for AWS connections, enabling users to perform data quality checks, summaries, and transformations on supported sources like Amazon S3. After updating to Airflow 3.3.1 and granting execution role access, customers can use the DuckDBExecuteQueryOperator without additional setup. This capability is available in all AWS Regions where Amazon MWAA is offered.
read more →

AWS launches GraphOver Datalakes with UI

🔍 AWS released an open-source local web UI, GraphOver Datalakes, that helps customers project data lake records into property graphs for analysis with Amazon Neptune Analytics. The tool consolidates projection, staging to Amazon S3, and import steps by running SQL/PGQ projection queries via Athena and previewing rows before ingest. It supports iceberg tables, works locally, and is available in all regions where Amazon Neptune Analytics is offered, with setup instructions provided.
read more →

Survey: Legacy Software Slows AI Modernization

🔍 A Google Cloud survey of 1,250 US IT and business decision-makers finds legacy, on-premises software acting as an "innovation tax" that slows AI adoption. Nearly half report delays or cancellations of AI initiatives due to maintaining legacy systems, with routine patching and maintenance consuming engineering capacity. Leaders call for interoperability and choice to avoid vendor lock-in and enable agentic AI.
read more →

Simplify cross-partition access with IAM outbound federation

🔒 AWS IAM outbound identity federation now supports cross-partition role exchanges so workloads can obtain a short-lived JWT in one partition and use sts:AssumeRoleWithWebIdentity to get temporary credentials in another partition. The post demonstrates setup and a walkthrough using S3 access across partitions, details required IAM policies and OIDC provider configuration, and highlights CloudTrail auditing and condition keys for restricting cross-partition trust. It also covers best practices, enforcement options, and compliance considerations.
read more →

DocumentDB Serverless Platform Version 4 Boosts Throughput

🚀 Amazon DocumentDB Serverless now delivers up to 35% higher throughput with the introduction of platform version 4. Platform versions represent managed improvements in performance, scaling, and features at the cluster level. New serverless clusters default to the latest platform version, while existing clusters on version 3 can be upgraded via a pending maintenance action. Upgrades are controlled by the customer and are irreversible; cluster version details are visible in the console or via the DescribeDBClusters API.
read more →

Athena adds managed system tables for usage data

📊 Amazon Athena now offers managed system tables that capture usage data automatically and make it queryable for observability, cost allocation, and audit purposes. When enabled, usage records appear within minutes and are stored as service-managed, read-only Apache Iceberg tables on Amazon S3 with configurable retention. Athena handles schema, partitioning, compaction, and snapshot maintenance, allowing teams to query usage via SQL without building custom ETL pipelines.
read more →

Safe Remediation: Reducing Exposure Without Disruption

🔒 Safe Remediation outlines how Check Point Exposure Management links prioritized vulnerabilities to validated production changes, applying controls through existing systems and verifying outcomes. It emphasizes evidence-based validation matched to the control type, using an agentless collection of telemetry and APIs. The process follows Observe, Prove, Act, and Confirm stages to protect production availability while reducing risk.
read more →

Google expands TPU portfolio with four eighth‑gen systems

🔷 Google announces two additional eighth-generation TPU systems — TPU 8tx for confidential inference and TPU 8i Max for massive context windows — expanding a lineup that already includes TPU 8t and TPU 8i. These systems combine modular silicon, custom hosts, and a low-latency network to address regulated workloads, large-memory reasoning, and high-density deployment. The company pairs the hardware with open-source software like TorchTPU, TPU Sync, and Cloud Pathways to simplify developer access and improve production performance.
read more →

Cloud Pathways distributed ML runtime reaches GA

🚀 Cloud Pathways is now generally available on Google Kubernetes Engine as a core runtime of the Google Cloud AI Hypercomputer. Originally developed by Google DeepMind to power frontier models like Gemini, Pathways provides a single-controller distributed ML runtime for large, heterogeneous TPU fleets with colocated Python sidecars and elastic fault tolerance. The release includes a Preview integration with Cluster Director for Slurm clusters and supports JAX-based multi-slice execution, resilient training, interactive notebooks, and shared long-running TPU pools.
read more →

GKE Dynamic Slicing for TPU Topology Agility

🔧 GKE Dynamic Slicing is now generally available, decoupling logical TPU slice formation from physical nodepool provisioning to accelerate AI workloads on TPUs. Dynamic slicing supports super-slicing to stitch TPU cubes into larger slices and sub-slicing to partition cubes for multi-tenant workloads, reducing startup times up to 5x and recovery times up to 4.5x while improving utilization and fault isolation.
read more →

TPU Sync: High‑Performance Model State for TPUs

🚀 TPU Sync is an open-source C++ library that unifies TPU HBM, host DRAM, shared storage, and the data center network into a single, co‑designed memory hierarchy to serve model state across TPU clusters. It uses layout‑preserving DMA and in‑place buffer aliasing to overlap transfers with computation, reducing KV transfer latency and Time to First Token by up to ~90% for very large models. TPU Sync integrates with PyTorch (TorchTPU) and JAX, supports disaggregated prefill/decode serving, and exposes low‑level primitives for experimental RL weight streaming and automated multi‑tier KV caching.
read more →

Cloudflare launches Vite Ecosystem Fund to support maintainers

💡 Cloudflare has begun distributing $301,300 from a $1 million commitment to the Vite Ecosystem Fund, supporting 44 projects and maintainers. The fund addresses accelerated maintainer burnout driven by AI-generated contributions and rising review workloads. Recipients were chosen by core Vite stack members, and eligible maintainers can receive stipends and hourly invoicing to sustain their work. Cloudflare will continue quarterly distributions alongside other open source support programs.
read more →

Amazon S3 Access Points Expand FSx Regional Support

🔔 You can now attach Amazon S3 Access Points to Amazon FSx for NetApp ONTAP in 4 additional AWS Regions and to Amazon FSx for OpenZFS in 20 additional Regions. An S3 Access Point for FSx provides an S3 endpoint so S3-based applications and services can access file data on FSx through the Amazon S3 API without copying or moving data. This enables integrations with AI/ML and analytics services like Amazon Bedrock, Amazon Athena, AWS Glue, serverless compute, and transfer services. The feature is now available across North America, South America, Europe, Asia Pacific, and Africa; create and attach access points via the console, AWS CLI, or AWS SDK.
read more →

AWS and Workday enable zero-copy analytics

🔗 AWS announces a bi-directional, zero-copy integration between AWS Analytics services and Workday Data Cloud, enabling customers to query across AWS and Workday datasets in place without replicating data. You can use Amazon SageMaker Unified Studio, Amazon EMR, AWS Glue, Amazon Redshift, and Amazon Athena to query Workday data via Workday Live Data Query or federate to Apache Iceberg tables through catalog federation. From Workday, register your AWS Glue Data Catalog Iceberg REST endpoint as an external catalog to query AWS Iceberg tables directly. The integration is available in all AWS commercial regions where AWS Glue is offered, with access governed by both AWS and Workday permissions.
read more →