Noisy Neighbor Detection - A New OSS Collector

Explore memory-related contention between containers through this 49-minute conference talk from the Linux Foundation's Open Source Summit. Discover why some requests experience significantly longer response times than others, with memory-related contention shown to increase latency by 4-13x, often triggered by garbage collection that existing observability tools cannot detect. Learn how current collectors only show high CPU utilization, leading to expensive scale-out solutions that don't address the core response time problem. Examine real-world patterns that trigger interference and survey detection methods for memory interference, including findings from production environments at Google, Alibaba, and Meta. Understand why traditional measurement approaches that sample every few seconds are inadequate for detecting the rapid transitions between intense resource competition and under-utilization that occur in modern server environments. Dive into the design of a new open source collector that combines CPU performance counters, eBPF, and high-resolution timers to identify noisy neighbors through millisecond-frequency measurements. Gain insights into future development directions and discover opportunities to contribute to this emerging observability solution.