High Availability With '503: Unavailable'

on YouTube ↗05m

" summary

Claims that embracing 503s as cheap, retryable failures and building apps with Envoy Istio and gRPC drastically reduces HA complexity while keeping user-facing success metrics. Implementation requires generated client libraries with retries-on-by-default, user experiences that tolerate a few quick retries, and idempotent APIs via per-request IDs plus server-side deduplication — Reboot uses its stateful framework to remember completed operations. Demonstrates fewer buffering components (no Knative queue‑proxy) and reports practical pitfalls and when this model is appropriate.

AI-written summary · about

" related talks

• 35mBuilding resilient applications with Durable ExecutionGiselle van Dongen · Devoxx Belgium 2024
• 20mUber's Failover Architecture: Reconciling Reliability and Efficiency in Hyperscale...Amit Kumar, David Bell, Mayank Bansal, Albert Greenberg, Sachin Sharma, Ivan Beschastnikh, Milind Chabbi, Srikanth Prodduturi, Kevin Xu, Ranjib Dey, Yufei Ren, Juan Marcano, Shriniket Kale, Subhav Pradhan, Miguel Covarrubias, Chien-Chih Liao, Sandeep Koushik Sheshadri, Wen Luo, Kai Song, Ashish Samant, Sahil Rihan, Nimish Sheth, Uday Kiran Medisetty, Kenneth Sejdenfaden Bøgh · NSDI '26
• 34mDistributed Systems - What Can Go Wrong and How to Get It RightFlorian Mair · Voxxed Days Ticino 2026
• 32mHow does your application behaves when everything goes wrong?Emerson Macedo · The Conf 2018
• 30mSturdy Clouds Creating Resilient Software Defined Data CentersChristian Huebner, Daniel Virassamy · OpenInfra Summit Berlin 2022
• 16mFrom System TAYE7 to High Availability: What, Why, How?Mouad El haous · BlaBlaConf 2025
• 36mDesigning for failure - architecting fault-tolerant systemJ. Derda · 4Developers 2015
• 30mBuilding Resilient Systems with StackingChris Keathley · ElixirConf EU 2019