GopherCon India talk - Go faster: optimizing Go programs

The article details how the authors diagnosed a significant memory regression introduced in Go 1.24 impacting their Kubernetes infrastructure. They used a combination of observability tools – including Prometheus, Grafana, and custom tracing – to pinpoint the issue to a change in the Go runtime's memory allocator, specifically related to large object handling. Ultimately, they collaborated with the Go team to identify and resolve the problem, demonstrating a robust approach to debugging performance regressions in distributed systems.