Consider the following general best practices when sizing your deployment:
- User density—If concurrency is a concern, be sure to calculate how many users will be using the environment at the peak of utilization. For example, if only 80 percent are using the environment at any time, the environment needs to support only that number of users (plus a failure capacity).
- Disaster recovery—For DR, Dell Technologies recommends implementing a dual/multisite solution. The goal is to keep the customer's environment online and, if there is an outage, to perform an environment recovery with minimum disruption to the business.
- Management and compute clusters—For small environments, using a combined management and compute cluster may be appropriate. For environments deployed at a larger scale, we recommend using separate management and compute layers. When creating a management cluster for a large-scale deployment, consider using R6515 servers to reduce the data center footprint. With a more flexible platform that accommodates a wider variety of VDI application workloads, R7525 servers are preferred for compute clusters.
- Network isolation—This design illustrates a two-NIC configuration per appliance with all the traffic separated logically using VLAN. When designing for larger-scale deployments, consider physically separating the host management and VDI traffic from the vSAN traffic for traffic isolation and to improve network performance and scalability.
- FTT—Dell Technologies recommends sizing storage with NumberOfFailuresToTolerate (FTT) set to 1, which means that you must double the amount of total storage to accommodate the mirroring of each VMDK. Depending on the scale of the environment, it may make sense to increase the FTT to 2. Consider the design of your VDI deployment to determine if the extra availability will outweigh the capacity overhead of increasing FTT.
- Slack space—Dell Technologies recommends adding an additional 30 percent of slack space to prevent automatic rebalancing of storage, which impacts performance. Automatic balancing occurs when the storage reaches 80 percent of the full threshold. Therefore, 70 percent is recommended to reserve a 10 percent buffer.
- All-flash compared with hybrid:
- Hybrid and all-flash configurations have similar performance results. Because hybrid uses spinning drives, consider the durability of the disks.
- Only all-flash configurations offer deduplication and compression for vSAN. Dell Technologies recommends all-flash configurations for simplified data management.
- All-flash configurations need considerably less storage capacity than hybrid configurations to produce similar FTT, as shown in the following table:
Table 11. Storage capacity comparison of all-flash and hybrid configurations VM size Failure tolerance method (FTM) FTT Overhead Configuration Capacity required Hosts required 50 GB RAID-1 (mirrored) 1 2x Hybrid 100 GB 3 50 GB RAID-5 (3+1) (erasure coding) 1 1.33x All-flash 66.5 GB 4 50 GB RAID-1 (mirrored) 2 3x Hybrid 150 GB 4 50 GB RAID-6 (4+2) (erasure coding) 2 1.5x All-flash 75 GB 6