Title

AWS re:Invent 2023 - A consistent approach to resilience analysis for critical workloads (ARC313)

Summary

John Fermento introduced the session and highlighted the importance of resilience in systems, drawing an analogy with the historical example of bullet hole analysis on WWII aircraft.
AWS released a resilience lifecycle framework, which is detailed in a white paper accessible via a QR code.
The session focused on the design, implement, and operate stages of the resilience lifecycle.
The resilience analysis framework (RAF) was introduced, which helps in assessing and improving the resilience of workloads.
RAF is based on the SEAMS model, which stands for Shared fate, Excessive load, Excessive latency, Misconfiguration, Bugs, and Single point of failure.
The RAF process involves understanding the workload, identifying critical user stories, and analyzing potential failure modes.
Trade-offs in resilience strategies were discussed, including cost, effort, complexity, operational burden, and consistency vs. latency.
The RAF process also involves assessing the likelihood and impact of potential failures and deciding on preventative or corrective mitigations.
Real-world customer examples were presented by Mike Haken, demonstrating how RAF was applied to address resilience issues.
Mike Golovnik shared insights on implementing RAF within AWS, emphasizing the need for proactive resilience thinking and integration with operational and development processes.

The RAF provides a structured approach to proactively analyze and improve the resilience of systems, which is crucial for mission-critical applications.
The SEAMS model within RAF helps identify common failure modes that can compromise resilience properties such as fault isolation, sufficient capacity, timely output, correct output, and redundancy.
Understanding the trade-offs involved in resilience strategies is essential for making informed decisions about where to invest resources for maximum impact.
Real-world examples highlighted the practical application of RAF and the benefits of patterns like constant work, hedging, and fault-isolated deployments.
RAF requires executive support, dedicated resources, and a team with a deep understanding of the system and a passion for resilience.
RAF is not just a technical process but also a cultural shift towards a resilience mindset, which involves continuous learning and improvement.
The RAF process can lead to significant improvements in system resilience by identifying and addressing risks that might otherwise be overlooked.
Integrating RAF with operational and feature development processes ensures that resilience analysis is not an isolated activity but part of the ongoing lifecycle of the application.