Speaker
Abstract
High-demand events can cause sudden traffic spikes that overwhelm even well-designed systems. In ticketing platforms, millions of users — alongside increasingly sophisticated automated agents — may arrive simultaneously, placing extreme pressure on backend services.
At SeatGeek, we observed that even elastic infrastructure has limits: autoscaling takes time to react, and systems must survive while capacity catches up. To address this gap, we designed a layered shielding architecture that distributes defensive responsibilities across multiple parts of the platform.
At the edge, caching, shielding, and admission control mechanisms such as queueing absorb traffic bursts before they reach the origin. API gateways enforce fairness through rate limiting and request validation. Deeper in the stack, Kubernetes-native networking policies and platform controls help contain failures and protect service boundaries.
This layered approach allows the system to shed load early, protect critical services, and degrade gracefully during extreme demand. But resilience is not static: traffic patterns evolve, new bottlenecks emerge, and systems must continuously adapt through observability and feedback signals.
In this talk, we will explore the architecture and operational lessons behind building multi-layer shields that protect core systems under internet-scale traffic, and share practical insights for designing resilient platforms that can withstand traffic stampedes without bringing down the entire ecosystem.
Interview
This talk explores how to design resilient systems that can withstand extreme traffic spikes without collapsing. Using real-world examples from ticketing platforms, I will show how distributing defensive responsibilities across layers — edge, gateway, and platform infrastructure — helps protect critical services during sudden demand surges. Senior engineers often operate systems where scaling alone is not enough; resilience requires intentional architecture and operational controls. The session focuses on practical patterns that help systems degrade gracefully rather than fail catastrophically.
Traffic patterns are becoming less predictable as automated agents, AI-driven clients, and global user demand increase system pressure. At the same time, modern platforms rely on complex distributed architectures in which small failures can quickly cascade. Leaders need to design systems that assume sudden spikes and evolving traffic behavior. Building resilience through layered defenses and clear operational signals is becoming essential for maintaining reliability at scale.
A common misconception is that cloud elasticity alone solves scalability problems. In reality, autoscaling takes time to react, and systems often experience instability before capacity catches up. Teams also struggle to identify truly critical services, manage noisy-neighbor effects in shared infrastructure, and detect early signals of system stress. Designing architectures that shed load early and protect the core system requires coordination across multiple platform layers.
I hope attendees rethink where traffic control happens in their systems. Instead of relying solely on backend scaling, they should introduce earlier defenses — such as caching, admission control, and rate limiting — to absorb pressure before it reaches core services. Even small changes at the edge or gateway layer can dramatically improve system stability during traffic spikes.
QCon focuses on real engineering experience rather than hype or vendor-driven content. Speakers share lessons learned from operating large-scale systems in production, including the trade-offs and failures behind architectural decisions. This creates an environment where senior engineers can learn from peers facing similar challenges. The emphasis on practical insight and honest technical discussion makes QCon particularly valuable.
QCon London 2026 is a three day conference for senior software engineers, architects and team leads. An international program committee of working engineers selects every session. Patterns and practices, not products and pitches.
Part of the track
Architecting for Resilience Hosted by Jonathan Magen Computer Scientist, Distributed Systems Specialist, 20+ Years in Software DevelopmentFrom the same track
Wednesday 18 March
10:35 Whittle (3rd Fl.) Session resilience How to Find Resilience Bugs in Systems that Don't Exist Hillel Wayne Author of "Logic for Programmers" and "Learn TLA+" Building correct distributed systems takes thinking outside the box, and the fastest way to do that is to think inside a different box. One different box is "formal methods", the discipline of mathematically verifying software and systems. 11:45 Whittle (3rd Fl.) Session decentralized Spritely: Infrastructure for the Future of the Internet Christine Lemmer-Webber, David Thompson Let's take back the internet! Learn about Spritely's work to re-decentralize the net with new foundational technologies that put users in control. 13:35 Churchill (Ground Fl.) Session architecture Understanding Progressive Collapse: How To Avoid A Cascading Failure Sam Newman Microservice, Cloud, CI/CD Expert, Author of "Building Microservices" and "Monolith to Microservices", 20+ Years Experience as a Developer Small things going wrong can quickly snowball. The cascading failure is often a nightmare scenario for any system. An initial problem, which in isolation seems like such a minor problem, can kick off a chain reaction of ever-increasing failures, potentially leading to catastrophic results. 14:45 Churchill (Ground Fl.) Session Keeping the Nation On-Air: How We Think About Resilience at the BBC Tom Everest Head of Department for Architecture and Supply Chain @BBC At the heart of the BBC is delivering value to all, serving audiences across the UK and the world on TV, radio, and online with trusted and impartial news and high-quality British content. 15:55 Churchill (Ground Fl.) Session Shielding the Core: Architecting Resilience with Multi-Layer Defenses Anderson Parra Staff Software Engineer @SeatGeek High-demand events can cause sudden traffic spikes that overwhelm even well-designed systems. In ticketing platforms, millions of users — alongside increasingly sophisticated automated agents — may arrive simultaneously, placing extreme pressure on backend services.