Disk drives in VxRail hosts are organized into disk groups, and disk groups contribute storage to the vSAN cluster. In a VxRail system, a disk group contains a maximum of one flash-cache drive and six capacity devices. Depending on the model, each VxRail node can be configured with up to four disk groups. The following figure shows the number of disk groups per node and the number of drives per disk group for each VxRail model.
Figure 42. Disk group configurations for VxRail based on latest generation servers
The disk group layout, based on capacity, performance, and availability, is determined when the system is designed and sized. Below are the considerations for designing disk-group layout:
- The number of SSD cache drives. Each disk group requires one and only one high-endurance SSD flash-cache drive. More disk groups require more SSD flash-cache drives. Currently, the VxRail system is offered with 400GB, 800GB, and 1600GB cache flash devices. For VxRail, the SSD cache drive must be installed in designated slots. See product documentation for specific slot locations.
- The total number of capacity drives available. Each disk group requires at least one capacity drive. Only capacity drive counts when determining the total capacity available for a vSAN Cluster.
- Cache requirements. Cache improves read and write performance in a hybrid storage configuration and write performance and endurance in an all-flash configuration. The optimal amount of cache depends on the active working set size. Workloads that have a larger working set size may require more cache, and this can be configured by either using larger SSD cache drives or configuring multiple disk groups with each disk group having one cache drive.
Note: All-flash configurations support a maximum of 600GB of write cache per disk group. Consider multiple disk groups for larger write cache per node.
- Performance. For the same total capacity, a larger number of smaller drives will provide more IOPs than fewer larger drives. Depending on the model, configuring more drives may require multiple disk groups.
- Fault Domains and recoverability. If the SSD cache drive fails, the full disk group goes offline and must be recovered. Larger disk groups take longer to recover. A node with only a single disk group will not be contributing any IOPs to the cluster during recovery, and this can reduce cluster performance.