AWS Updates 2026-11-19 | Cloud Provider News

Cut idle EC2 spend faster with Cost Optimization Hub

AWS just made it easier to act on idle compute waste with a new Cost Optimization Hub update. Cost Optimization Hub can now stop idle Amazon EC2 instances directly from the recommendations page, so you don’t have to jump through extra console steps. That makes it quicker to turn recommendations into action and helps teams reduce unnecessary compute costs across accounts and Regions.
It’s a small workflow change, but it matters when you’re trying to move fast on waste. By shortening the path from insight to action, teams can spend less time chasing down idle instances and more time focusing on higher-value optimization work.

Run Lambda workloads more efficiently when demand drops

AWS Lambda now gives managed instances a cleaner way to scale down when they’re not busy. Lambda Managed Instances can now scale execution environments to zero during idle periods, then automatically scale back up when requests come back.
That directly reduces idle compute spend for intermittent workloads, which is especially useful for async processing and dev/test environments. You still keep Lambda’s operational simplicity, but with less paying for capacity you aren’t using.

Lower Kafka costs with M8g-backed MSK brokers

Amazon MSK just got a cost-focused upgrade for teams running Kafka on AWS. MSK Express Brokers now support Graviton4-based M8g instances, and AWS says that cuts compute costs by 15%, cross-AZ transfer charges by 50%, and delivers up to 30% total cost savings versus M7g.
That’s a meaningful improvement for managed Kafka workloads because it reduces both compute and data-transfer spend. For teams watching streaming costs closely, this one has a very direct FinOps payoff.

Simplify real-time processing with cheaper Kinesis queues

Amazon Kinesis Data Streams now has a simpler way to consume data and cut costs. Kinesis Data Streams now supports queues as a consumption model, which can reduce stream processing costs by up to 43% compared with the Kinesis Client Library.
It also removes the need to run a DynamoDB table or custom metrics, which lowers operational overhead along with compute use. That makes it easier to run real-time pipelines without carrying as much extra infrastructure around them.

Get better read performance without the usual cache overhead

Amazon DynamoDB caching now works with ElastiCache Serverless for Valkey. The new setup provides sub-millisecond reads without managing DAX nodes or custom cache infrastructure.
For teams with frequently accessed DynamoDB data, that means a simpler caching path and less operational work to maintain. It can also improve read efficiency, which helps keep spend focused on the traffic that actually matters.

Use Spark tuning guidance to trim EMR Serverless costs

Amazon EMR Serverless now includes a built-in way to look for Spark tuning wins after jobs finish. The Apache Spark optimization agent analyzes completed jobs and recommends cost- or performance-focused tuning changes.
That’s useful when you want concrete guidance instead of guessing where waste is happening. It can help teams lower Spark spend and improve workload efficiency using telemetry-backed recommendations.

Put Apache Spark, Iceberg, and Delta Lake on a fresher EMR release

Amazon EMR 8.2 is now available as a new LTS release. The release includes Spark 4.1.3, Iceberg 1.11.0, and Delta Lake 4.2.0, along with performance and governance improvements across EMR on EC2, EKS, and Serverless.
For FinOps teams, platform upgrades like this can matter because better performance and smoother governance can reduce operational friction at scale. If you’re already on EMR, this gives you a newer base to run workloads on without changing the service model.

Improve Kafka broker efficiency on standard M8g instances

Amazon MSK Standard Brokers now support Graviton4-based M8g instances. AWS says this brings up to 35% lower P99 produce latency and better consumer catch-up performance.
That’s a nice fit for Kafka teams looking for more efficient broker capacity without giving up performance. Better latency and catch-up behavior can help workloads run more smoothly while keeping infrastructure choices more efficient.

Review EC2 policies carefully before enforcing cost and network controls

AWS Organizations has a detailed EC2 policy guide with examples for shared account controls. The page explains EC2 policy syntax, supported policies, and how settings like VPC Block Public Access and Networking Flexible Cost Allocation work.
This is most useful when you’re planning governance changes across accounts, because noncompliant actions will fail and account admins can’t override the enforced settings locally. It’s a reminder that centralized controls can help with consistency, but they also need careful rollout so you don’t break important workflows.

FinOps Weekly
FinOps Weekly
Articles: 296