Posted On: Mar 4, 2022

You can now customize how data is stored on your Amazon FSx for OpenZFS file system to optimize performance for database applications and other workloads that consistently read and write data in small increments.

Amazon FSx for OpenZFS provides fully managed cost-effective shared file storage powered by the popular OpenZFS file system. The ZFS “record-size” setting informs how your data is laid out as it is written to your file system’s disks, specifically the size of individual chunks or “blocks" used to store it. The default FSx for OpenZFS record-size is automatically tuned to serve the vast majority of workloads, but some workloads that consistently read or write data in small chunks can benefit from a custom record-size. Aligning the ZFS record-size to your workload’s read and write patterns (e.g., the page size on a database application) minimizes the amount of data transferred to support each individual operation, optimizing your realized file system throughput, latency, and IOPS by up to 3x.

Starting today, you can customize the record-size on your FSx for OpenZFS file system via the AWS Console, AWS CLI, or FSx API, and at no additional cost. This feature is supported on all new or existing file systems and in all regions where FSx for OpenZFS is available. See the Amazon FSx documentation for more information.