How to use FSx for OpenZFS metrics - FSx for OpenZFS
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

How to use FSx for OpenZFS metrics

There are two primary architectural components of each Amazon FSx file system:

  • The file server that serves data to clients that access the file system.

  • The storage volumes that host the data in your file system.

FSx for OpenZFS reports metrics in CloudWatch that track performance and resource utilization for your file system's file server and storage volumes. The following diagram illustrates an Amazon FSx file system with its architectural components, and the performance and resource CloudWatch metrics that are available for monitoring. The key property for a set of metrics is the file system property that determines the capacity for those metrics. Adjusting that property modifies the file system's performance for that set of metrics.

Diagram displaying the different types of FSx for OpenZFS Cloudwatch metrics.

You can use the Monitoring & performance panel on your file system's dashboard in the Amazon FSx console to view the metrics that are described in the following table. For more information, see Accessing CloudWatch metrics.

Monitoring &performance panel How do I... Chart Relevant metrics
Summary

...determine my file system's total throughput?

 Total throughput

SUM(DataReadBytes + DataWriteBytes)/Period (in seconds)

...determine my file system's total IOPS?

 Total IOPS

SUM(DataReadOperations + DataWriteOperations + MetadataOperations)/Period (in seconds)
...determine the number of connections that are established between clients and the file server? Client connections ClientConnections
Storage ...determine how much primary storage is available? Available primary storage capacity (bytes)

StorageCapacity {SSD} - UsedStorageCapacity {SSD}
...determine the percentage of used primary storage for my file system? Primary storage capacity utilization (percent)

StorageCapacity {SSD} * 100/UsedStorageCapacity {SSD}
File server performance ...determine the network throughput for clients that access the file system as a percentage of the file system's provisioned throughput? Network throughput utilization NetworkThroughputUtilization
...determine the disk throughput between the file server and its storage volumes as a percentage of the provisioned limit, which is determined by throughput capacity? Disk throughput utilization FileServerDiskThroughputUtilization
...determine the percentage of available burst credits for disk throughput between the file server and its storage volumes? Disk throughput burst balance FileServerDiskThroughputBalance
...determine the amount of disk IOPS between the file server and storage volumes as a percentage of the provisioned limit, which is determined by throughput capacity? Disk IOPS utilization FileServerDiskIopsUtilization
...determine the percentage of available burst credits for disk IOPS between the file server and storage volumes? Disk IOPS burst balance FileServerDiskIopsBalance
...determine the file server's CPU utilization percentage? CPU utilization CPUUtilization
...determine the file server's memory utilization percentage? Memory utilization MemoryUtilization
...determine my workload's usage of the file system's in-memory (ARC) and NVMe (L2ARC) caches? Cache hit ratio FileServerCacheHitRatio
Disk performance ...determine the IOPS for operations that access storage volumes as a percentage of the provisioned limit determined by SSD storage capacity? Disk IOPS utilization (SSD) DiskIopsUtilization
...determine the savings from data compression? Compression ratio (UsedStorageCapacity/CompressionRatio) - UsedStorageCapacity
Note

We recommend that you provision throughput capacity so that the utilization of any performance-related dimension,—such as typical network, CPU, and memory utilization is less than 50%. This ensures that you have enough spare throughput capacity for unexpected spikes in your workload, as well as for background storage operations.