- cloudengineer.ing
- Posts
- Sep 2 - Sep 8 Updates
Sep 2 - Sep 8 Updates
Last Week Cloud Platform Updates
Sep 2 - Sep 8 | AWS, Azure & GCP Updates
Amazon ECS now supports AWS Graviton-based compute with AWS Fargate Spot, allowing customers to run fault-tolerant Arm-based applications with up to 70% cost savings compared to standard Fargate prices. This integration leverages AWS Graviton processors, designed for optimal price-performance for cloud workloads. By running Arm-based workloads on AWS Fargate Spot, customers can achieve serverless simplicity while optimizing costs with Spot capacity. To use this feature, configure your task definition with cpu-architecture = ARM64 and select FARGATE_SPOT as the capacity provider.
Amazon S3 introduces the alpha release of Storage Browser for S3, an open-source component that integrates with web applications to provide a user-friendly interface for accessing S3 data. Available in AWS Amplify JavaScript and React client libraries, this tool allows authorized end users to browse, download, and upload data directly within your applications.
Amazon S3 Access Grants now includes the ListCallerAccessGrants API, which allows AWS IAM principals and AWS IAM Identity Center end users to list all S3 buckets, prefixes, and objects they have access to, as defined by their S3 Access Grants. This API enables customers to build applications that identify and act on data accessible to specific end users.
Amazon DynamoDB now supports Attribute-Based Access Control (ABAC) for tables and indexes, allowing you to define access permissions based on tags attached to users, roles, and AWS resources. With ABAC, you can use tag-based conditions to allow or deny specific actions, providing more granular access control aligned with your organizational structure.
AWS now allows configurable TCP idle timeout settings for Network Load Balancer (NLB) and Gateway Load Balancer (GWLB) to better handle long-lived traffic flows.
NLB: Configure TCP idle timeout from 60 to 6000 seconds, reducing retries and latency for applications like databases and streaming services. The default remains 350 seconds. Use the tcp.idle_timeout.seconds attribute for configuration.
GWLB: Supports configurable TCP idle timeout from 60 to 6000 seconds to maintain uninterrupted traffic flows for applications like financial systems. The default is 350 seconds, configurable via the tcp.idle_timeout.seconds attribute.
Azure Public IPs are zone redundant by default. Unless you specifically select a single zone when you create Standard Public IPs, it will be zone redundant. Zone-redundant by default for Public IPs has no extra cost associated with it. A zone-redundant IP is created in all 3 zones for a region and can survive any single zone failure improving the resiliency of your application using this public IP. This functionality is available in limited regions and will continue to expand.
Azure Bastion now supports Microsoft Entra ID authentication for SSH connections in the portal. With Microsoft Entra ID authentication, there are two main benefits to users connecting to their virtual machines. First, it eliminates the need for local authentication mechanisms, reducing a point of attack for malicious actors. Second, with Microsoft Entra ID authentication set as the authentication mechanism, instead of providing additional authentication to connect, users can experience a one-click sign-on into their virtual machines.
We are excited to announce new enhancements to our GenAI Gateway capabilities, specifically designed for large language model (LLM) use cases. Building on our initial release in May 2024, we are introducing new policies to support a wider range of LLMs via the Azure AI Model Inference API. These new policies offer the same robust functionality as our initial offerings but are now compatible with a broader array of models available in Azure AI Studio.
Azure Site Recovery has introduced automatic renewal of certificates for on-premises to Azure disaster recovery. Azure Site Recovery uses various components for doing disaster recovery (DR) from on-premises to Azure. Certificates are essential for communication between components and need regular renewal to avoid disruptions in Azure Site Recovery operations (like data replication, etc.). This new capability ensures automatic renewal of certificates without affecting ongoing replication.
We are excited to announce the Public Preview of Azure Container Storage enabled by Azure Arc Edge Volumes, a versatile new addition to our Azure storage solutions designed to enhance data handling at the edge. Edge Volumes offers two key capabilities: Local Shared Volume and Cloud Ingest Volume.
Metrics Management: The Cloud Monitoring Metrics management page now shows the sources of metric reads and allows you to exclude unnecessary metrics, reducing ingestion costs.
GPU Observability: The Observability tabs on the Compute Engine VM instances page now include charts for GPU metrics from the NVIDIA Management Library (NVML). These charts are available for VM instances with GPUs, the Ops Agent, and the NVIDIA GPU driver installed. Access fleetwide or per-instance GPU metrics through the Compute Engine > VM instances > Observability section.
Firestore now offers K-nearest neighbor (KNN) vector search capabilities, allowing you to perform vector searches with inequality filters, retrieve calculated vector distances, and set distance thresholds. This feature is now generally available (GA).
BigQuery now supports vector search and vector index features, allowing you to use the VECTOR_SEARCH function to find semantically similar entities in embeddings. Vector indexes can be used to make vector searches more efficient, with a trade-off of returning more approximate results. These features are now generally available (GA).
Cloud SQL for MySQL: You can now specify a preferred zone when cloning a zonal instance and set preferred primary and secondary zones for a regional instance. This ensures Cloud SQL can assign these preferred zones to prevent downtime if the current zones go down.
Cloud SQL for PostgreSQL: Point-in-time recovery (PITR) now supports restoring a zonal instance to a preferred primary zone and a regional instance to both preferred primary and secondary zones. Additionally, you can specify preferred zones when cloning instances to prevent downtime.
Cloud SQL for SQL Server: Similar to MySQL, you can now specify preferred zones when cloning zonal or regional instances to avoid downtime if the assigned zones go down.