Design Factor | Detail |
Controller Queue Depth | Controller queue depth impacts the rebuild/resync times. A low controller queue depth may impact the availability of production virtual machines during rebuild/resync. A minimum queue depth of 256 is required for vSAN. |
Number of disk groups | • The number of disk groups impacts fault isolation as well as rebuild/resync times. • Fault isolation: Configuring more than 1 disk group allows better tolerance against SSD failures, because data is spread across more disk groups. • Rebuild/resync times: Configuring more than 1 disk group allows faster rebuilds/resyncs. |
Number of hard drives (HDDs) in a disk group | The number of HDDs in a disk group has an impact on the performance of vSAN. While a single HDD is the minimum requirement for a disk group, for better performance when there are more virtual machines, and better handling of rebuild/resync activities, VMware recommends configuring more than 1 HDD per SSD, per the guidance above. |
Class of SSDs | The class of SSD you choose has a direct impact on the performance of your overall system. |
Balanced compared with unbalanced cluster | An unbalanced cluster can impact vSAN performance, as well as the rebuild/resync times. A balanced cluster delivers more predictable performance, even during hardware failures. In addition, performance impact during resync/rebuild is minimal when the cluster is balanced. |
1-GB vs 10-G Ethernet | The choice of 1-GbE vs 10-GbE Ethernet has an impact on the vSAN performance. For larger, higher performing and production workloads, 10 GbE is recommended. |