eBooks/eGuides

34 Another thing to keep in mind is that if you select a CPU ratio in the middle of those ranges, it will provide the freedom to scale the consolidation density upwards should the environment continue to perform within tolerances. One thing to note is that there is no place to configure these CPU ratios as a setting in any other tools today. These are attributes that must be declared in the design and become data points that one will need to account for in the management and scaling of the environment. Just as much as memory and clock speed, the CPU ratio needs to be calculated into the decision to add more VMs to a cluster, and when to add another host to a cluster to provide more resources. One can manage the CPU ratio through manual calculations by gathering data. Some administrators use a PowerShell script that will gather data and present the ratio as the output from the script. With a script, it could run as a scheduled job daily to ensure one is not violating the ratio and be in danger on any of the clusters. The RAM or memory bus frequency is also associated with the compute sizing. The rule of thumb when sizing memory is to aim for the highest density with the fastest bus speed budgets will allow. The challenge often faced with memory is that slower memory can result in idle CPU cycles waiting for read/write transactions to RAM to complete. Incorporating GPUs into your VDI clusters and design will typically affect the density of users per host also. This is directly related to the number of GPU cards and number of GPUs per card that can be placed in your desired host and then the vGPU profile selected for your users. Simple example of a host that can accept two GPU cards, each with one GPU. Then a vGPU profile that allows 16 users per card is chosen, this means that only 32 users that need GPU would fit on that type of host. If there is available CPU and memory on the host it can still run non-GPU VMs. Identifying the correct vGPU profile is important for efficient sizing.

• Cover