AWS has quietly overhauled core services - EC2, S3, networking, Lambda and more - so old best practices are often wrong today.
Technical leaders still hear the same old AWS myths, but the platform has quietly rewritten its foundations. The article pulls together the most surprising changes across services so you can stop relying on outdated assumptions and make decisions that reflect the modern reality of the cloud.
EC2 now lets you swap security groups and IAM roles on a running instance, resize or detach EBS volumes without shutdown, and even force a hard stop or termination. Live-migration to new hosts has reduced instance degradation notices, spot markets have smoothed into predictable price curves, and dedicated instances are no longer a compliance requirement. New accounts default to AMI Block Public Access, and the default security posture has hardened across the board.
S3 stopped being eventually consistent and now offers read-after-write guarantees, removing the need to randomize object prefixes. New buckets come with default encryption, Block Public Access, and disabled ACLs. Glacier is now just another storage class inside S3, with predictable pricing and dramatically faster restores, eliminating the old "expensive and slow" narrative.
Networking, Lambda, and storage have also caught up. VPC Classic is gone, IP address pricing is transparent, and newer options like Transit Gateway, VPC sharing and Lattice simplify cross-account connectivity. Lambda supports 15-minute timeouts, Docker images, 10 GB RAM, and 10 GB /tmp, while VPC cold starts have been tamed. EBS volumes now deliver full performance out of the box, and snapshot-driven boots can be accelerated with a full-disk read.
Cost-management tools have matured: Reserved Instances are being replaced by Savings Plans, EC2 billing is now per-second, and the Cost Anomaly Detector is free. IAM roles should be the primary permission model, with Identity Center replacing the old SSO, and multi-MFA on root accounts is now supported. These concrete shifts mean leaders can cut waste, avoid legacy traps, and align teams around a cloud that behaves predictably today.
Check out the full stdlib collection for more frameworks, templates, and guides to accelerate your technical leadership journey.