CloudFront Flat-Rate Plans Are Now Easier to Automate

Looking for more predictable delivery costs? Amazon CloudFront now lets you manage flat-rate pricing plans through APIs, the AWS CLI, SDKs, CloudFormation, and the AWS CDK.

The plans bundle content delivery, AWS WAF, DDoS protection, DNS, logging, and edge compute under a fixed monthly price. The new management options make it easier to include plan selection and changes in automated workflows.

That combination can improve spend predictability while reducing manual work for teams managing CloudFront configurations at scale.

AWS Marketplace Private Offers Can Renew Automatically

Renewing Marketplace contracts just got more predictable. AWS Marketplace now supports automatic renewals for private offers, carrying forward negotiated pricing and contract terms.

Teams can configure price-change rules, and buyers have controls to opt out of renewals. This gives procurement and FinOps teams more control over how renewal costs are handled.

The update can also reduce the risk of unplanned service lapses and unmanaged renewal spending.

Graviton5 C9g and C9gd Instances Reach Tokyo

Teams running compute-optimized workloads in Japan now have another instance option. Amazon EC2 C9g and C9gd instances powered by Graviton5 are available in the Asia Pacific (Tokyo) Region.

The instances offer up to 25% better compute performance than Graviton4-based instances. They’re available through Savings Plans, Spot, On-Demand, and dedicated purchasing models.

That range of purchasing options gives FinOps teams more flexibility when placing workloads and managing compute commitments.

Graviton5 M9g and M9gd Instances Expand Across Four Regions

Memory-intensive workloads have more Graviton5 placement options. Amazon EC2 M9g and M9gd instances are now available in Ireland, Singapore, Sydney, and Tokyo.

The instances provide up to 25% better performance than Graviton4-based instances and are positioned for improved price-performance and energy efficiency.

That can give teams additional options when reviewing instance choices for memory-intensive workloads across these regions.

SageMaker Feature Store Supports Lower-Cost Feature-Level Writes

High-volume machine learning pipelines can now update less data per write. Amazon SageMaker Feature Store lets teams update individual features within a record without rewriting the entire record.

This lowers write latency and write costs, particularly when multiple processes update different features in the same record.

The change can make Feature Store usage more efficient for workloads with frequent, targeted updates.

CloudWatch Database Insights Adds Self-Managed PostgreSQL on EC2

Database visibility now extends beyond managed database services. Amazon CloudWatch Database Insights can monitor self-managed PostgreSQL databases running on Amazon EC2, alongside Amazon RDS and Aurora.

Teams get fleet-wide views of database load, wait events, queries, and host metrics. These details can help identify performance bottlenecks and potential overprovisioning.

With that information, teams can investigate whether database capacity is aligned with workload needs.

Redshift RG.Large Supports Single-Node Clusters at a Lower Price per vCPU

Smaller Amazon Redshift workloads now have a single-node option. Redshift RG.Large instances support single-node clusters for proofs of concept, testing, and workloads that don’t require high availability.

RG instances provide up to 2.4 times the performance of previous-generation RA3 instances at 30% lower price per vCPU.

That gives teams a lower-cost configuration for suitable smaller-scale workloads and non-high-availability use cases.

Aurora MySQL Adds Multi-Source and Delayed Replication

Aurora MySQL now supports more flexible replication designs. Amazon Aurora MySQL can consolidate multiple source databases into one replica and use intentionally delayed replication.

These capabilities can simplify reporting, backups, shard consolidation, and recovery workflows. They can also reduce the need for separate replication infrastructure.

That may help teams simplify supporting systems while using database capacity more efficiently.

SageMaker Batch Transform Adds G6e GPU Instances

Offline inference workloads can now use G6e instances with NVIDIA L40S GPUs. Amazon SageMaker Batch Transform supports G6e instances for batch inference.

Using batch processing instead of persistent endpoints can reduce idle infrastructure costs for large datasets and intermittent GPU-intensive workloads.