AWS Updates 2026-10-22 | Cloud Provider News

AWS

Bedrock prompt shortening can trim AI input costs

Amazon Bedrock Advanced Prompt Optimization now supports prompt shortening, reducing prompt length while keeping or improving evaluation scores. Shorter prompts can lower input token costs and improve latency.

For teams running generative AI workloads, that’s a pretty direct FinOps win. Less input text means less to pay for, and faster responses can also make workflows feel smoother.

Lambda now helps stop runaway event costs on EventBridge

Lambda recursive loop detection now covers Amazon EventBridge custom event buses, which helps prevent runaway invocations and unexpected charges. This update is especially relevant for FinOps because misconfigured event-driven workloads can rack up costs fast.

The main benefit here is simpler cost protection. If a recursive loop starts, this expanded detection can help stop it before it turns into a bigger billing problem.

ECS Gateway could make shared ingress a lot simpler

Amazon ECS Gateway is a new managed ingress option for ECS that dynamically provisions and manages Application Load Balancers and related resources. It lets up to 25 services share a single ALB, which can help cut infrastructure sprawl.

That’s useful if you’re trying to keep network setup lean while still supporting multiple services. Fewer load balancers can also mean less operational overhead to manage across clusters.

CloudWatch log centralization expands for broader monitoring

CloudWatch log centralization is now available in AWS GovCloud (US) and 13 additional commercial Regions. The feature consolidates logs across accounts and Regions into a single destination.

That makes monitoring and governance easier at scale. When logs live in one place, it’s simpler to track what’s happening across a larger AWS footprint.

CloudWatch Omni reaches more Regions

CloudWatch Omni is now available in eight additional AWS Regions, expanding an AI-powered observability experience that spans AWS accounts, Regions, and even other clouds.

This can make it easier to spot issues faster without jumping between too many tools or views. That saves time for operations teams and can help reduce waste from slow-to-detect problems.

Redshift keeps system table data longer without custom pipelines

Amazon Redshift now extends long-term system table retention via S3 Tables, letting customers retain system table data beyond seven days in additional Regions. It also avoids the need for custom ETL pipelines.

That’s helpful when you want better observability and auditing without building extra data movement jobs. It can make it easier to study warehouse performance and usage over time.

Time Sync gets more accurate across more EC2 instances

Amazon Time Sync Service now supports microsecond-accurate time on 135 additional EC2 instance types across all commercial Regions. This supports distributed systems and other time-sensitive workloads.

More accurate time can help with performance and observability, especially when you’re dealing with systems where ordering and timing matter.

DynamoDB global tables get more flexible replica handling

Amazon DynamoDB global tables now support disassociating replicas without deleting the table or data. That gives teams more flexibility to stop replication for a Region while continuing to use the table locally.

This can reduce operational overhead when you no longer need a replica in a specific place. It also helps with cost control by making replication easier to adjust.

FinOps Weekly
FinOps Weekly
Articles: 289