Mountpoint for Amazon S3 adds memory usage controls

Mountpoint for Amazon S3 now supports configurable memory limits, addressing issues where its memory consumption could grow over time and disrupt other applications. Users can either specify a fixed memory target or enable automatic adjustment based on the runtime environment. This change is designed to prevent performance degradation in shared-memory workloads, such as machine learning training or analytics. The feature is available in all AWS Regions and applies to both standalone and containerized deployments.

Previously, Mountpoint’s memory usage was unbounded, which could lead to instability when running alongside memory-intensive applications. The new controls allow users to reserve memory for critical workloads by setting a hard limit or letting Mountpoint dynamically adjust based on available resources. In containerized environments like Amazon EKS, Mountpoint can automatically detect and respect the container’s allocated memory budget, ensuring predictable performance.

The update introduces two modes for memory management: a user-defined limit or an environment-aware default. When memory pressure occurs, Mountpoint throttles operations to stay within the specified budget, preventing out-of-memory errors. This is particularly useful for workloads with strict memory constraints, such as real-time analytics or batch processing pipelines. The feature is backward compatible and does not require changes to existing configurations.

To adopt the new memory controls, users can upgrade to the latest version of Mountpoint via the GitHub repository. AWS provides documentation on configuration options, including examples for setting memory limits and integrating with containerized environments. The feature is available globally, and no additional AWS services or permissions are required to enable it.