Breaking Strain: A Story About Capacities and Testing

Our systems seldom fail all at once. Instead, an accumulation of failures adds up into a disaster, or a collection of latencies becomes an outage. What can we do to build systems that are more robust than our outage budgets? When do we need to decide to deliver less instead of nothing? And how can we possibly test our solutions under realistic conditions? Perfectly optimized systems are extremely brittle -- we can see that now in a lot of our physical world. Systems operating near the limits of their capacity can be slow or intermittent. But systems with too much excess capacity are expensive. Finding the right balance of capacity needs and wiggle room means that organizations need to understand what they require and be able to change those requirements on the fly. How much and how fast can your system change to avoid breaking, or to take advantage of new opportunities? Audiences will leave with new questions to ask about testing in production, automatic traffic management, application delivery, and modular designs.