RESIN is a memory leak detection service designed to address memory leaks in large cloud infrastructures. Memory leaks impact performance, stability, and user experience in cloud environments. Traditional detection methods are not suitable for cloud environments due to accuracy and scalability issues. RESIN uses a two-tier detection scheme to identify memory leak symptoms at both the component and process levels.
Once a memory leak is detected, RESIN takes a snapshot of the active heap to analyze the root cause of the leak. It leverages the Windows heap manager for live profiling but minimizes overhead by prioritizing host selection and using trigger-based strategies. When a memory leak is detected, RESIN automatically attempts to mitigate the issue, taking actions such as restarting processes or services to minimize impact.
RESIN has been running in production in Azure since late 2018, monitoring millions of host nodes daily with an 85% precision and 91% recall rate. It has significantly reduced unexpected virtual machine restarts and allocation errors, leading to improved reliability and performance of cloud infrastructure.
Overall, RESIN offers end-to-end memory leak detection capabilities to comprehensively address memory leaks in cloud environments, reducing the risk of serious outages caused by memory leaks.
Article Source
https://azure.microsoft.com/en-us/blog/advancing-memory-leak-detection-with-aiops-introducing-resin/