Avoiding the Pitfalls of Autoscaling with Constant Work

Explore an alternative approach to managing risks associated with auto-scaling, geo-redundancy, and high-availability for internet applications in this 36-minute conference talk from Strange Loop 2022. Delve into the concept of constant work as a strategy to mitigate regional outages and cascading failures, moving beyond traditional scaling and failover patterns. Examine systems implementing constant work, address the trade-offs between scale and cost, and identify potential areas for investment in this approach. Gain practical insights through a concrete example comparing both constant work and on-demand scaling methods. Presented by David Grizzanti, Principal Engineer at The New York Times, this talk offers valuable perspectives on improving infrastructure automation and engineering leadership for those interested in enhancing the resilience and efficiency of large-scale internet applications.