Season 5 Episode 19: CloudFront Pricing, Container Migrations, and AI Sovereignty

About this episode

In Season 5, Episode 19, Karl and Jon are joined by Anwaar Hussain, Cloud Infrastructure Architect at AWS, to discuss the latest AWS updates, including Amazon CloudFront Premium Flat Rate Pricing Plans, which offer more flexible tier options for request counts and data transfer rates; AWS Transform Containerization, an AI-powered capability for containerizing applications during migration to ECS or EKS; Amazon Q in QuickSight Requirements Analysis, which helps identify contradictions and improve code quality; AI Sovereignty on AWS, covering data residency, operator access restrictions, and responsible AI certification; and analysis of the May 2024 US-East-1 outage, which reinforces the importance of multi-AZ architecture and disaster recovery testing. Karl also questions why Jon did not spend his time off in lieu having a test nap in the camper van he wants to buy.

 

07:07 - Amazon CloudFront Premium Flat Rate Pricing Plans - Expanded Options 

AWS has enhanced CloudFront flat rate pricing with more flexible tier customization, letting customers choose combinations from 500 million to 6 billion requests and 50 to 600 TB of data transfer without manual AWS support. This helps mid-market enterprises access custom pricing, improve cost forecasting, and align cloud spending with fixed budget planning.

 

14:28 - AWS Transform Adds Containerization Capability During Migration 

AWS Transform now uses AI to automate application containerization, going beyond lift-and-shift migrations. It analyzes source code, creates Dockerfiles, builds and scans images, pushes them to ECR, and deploys to ECS or EKS. Supporting platforms such as IBM z/OS, Fujitsu GS21, VMware, and .NET, it helps organizations modernize faster while allowing consultants to focus on higher-value work like upskilling, architecture, and complex transformation.

 

23:33 - Amazon Q in QuickSight - Requirements Analysis Feature (GeekWire: "AWS Targets AI Slop") 

Amazon Q has launched a requirements analysis feature designed to improve code quality by identifying gaps and contradictions in application specifications before development begins. Using a neurosymbolic AI pipeline—combining neural networks with SMT (Satisfiability Modulo Theories) solvers—the tool translates requirements into logical format, searches for mathematical contradictions, and presents clarifying questions to users about edge cases, conflicting requirements, and use case priorities. This addresses a core challenge with AI-generated code: while AI excels at optimizing for high-throughput scenarios, real-world applications often have different constraints and lower usage patterns that require deliberate architectural choices. The feature was released alongside parallel task execution (allowing simultaneous sub-agent work on independent tasks) and quick plan mode (combining requirements analysis, design, and task listing). Combined with Q's existing spec mode, these tools reduce implementation time and improve code quality, though they emphasize that prompt engineering rigor and human oversight remain critical to avoiding poor-quality outputs in production environments.

 

36:49 - AWS Security Blog - Enabling AI Sovereignty on AWS 

Stefan Israel, lead of AWS’s European Sovereign Cloud, outlined how AWS addresses AI sovereignty through data residency, operator access controls, and cultural-linguistic model considerations. Building on AWS’s Digital Sovereignty Pledge, the guidance covers where models run, how data moves through AI workflows, and how responsible AI is governed. AWS’s ISO/IEC 42001 certification adds assurance for customers still cautious about generative and agentic AI adoption.

 

44:04 - Mashable - AWS Reveals Cause of May 2024 Outage

AWS disclosed that the May 2024 US-East-1 outage was caused by a cooling failure in a single data center within one availability zone. While only part of the region was affected, the incident reinforces the need for multi-AZ architecture, regular disaster recovery testing, and verified backup restoration plans, as many organizations still rely on single-AZ dependencies or untested recovery processes.