Sizing calculator

Use the following formulas to provide an estimated infrastructure for VM deployments. Considering the overhead and high availability, these estimates offer a close approximation for effective resource planning.

Expected input

Number of VMs
Total number of virtual machines planned.
VM Types
Are the VMs of the same type or different types with distinct vCPU, Memory, and Storage requirements?

Formulas

  • CPU Capacity: The following formula calculates the total vCPU capacity that is available for VMs. Subtract the overhead for control plane and other necessary services:
    Total Available (vCPUs) = (vCPUs_per_node × node_count − control_plane_requirements − overhead) × (1−ha_reserve_percent) × (overcommit_ratio)
    Parameter Description
    vCPUs_per_node The number of vCPUs per compute node. For example, with 64 cores and 2 threads per core, you would have 128 vCPUs per node.
    node_count The number of compute nodes.
    control_plane_requirements CPU is reserved for control plane functions.
    overhead Overhead for management services. For example, Data Foundation in IBM Storage Fusion.
    ha_reserve_percent Percentage of capacity reserved for high availability.
    overcommit_ratio Factor to consider when you overcommit CPU resources. For example, 4:1.

  • Memory Capacity: Similarly, the memory capacity calculation accounts for the overhead and reserves for high availability:

    Total Available Memory (GiB) = (total_node_memory × node_count − control_plane_requirements − overhead) × (1 − ha_reserve_percent)
    Parameter Description
    total_node_memory Total physical memory per compute node.
    control_plane_requirements Memory reserved for control plane operations.
    overhead Overhead for management services. For example, Data Foundation in IBM Storage Fusion.
    node_count Number of compute nodes.
    ha_reserve_percent Percentage of memory reserved for high availability.
  • Storage Capacity: Storage capacity is calculated after accounting for the overhead needed for management and other services:
    Total Available Storage (TiB) = total_storage_capacity − storage_overhead
    Parameter Description
    total_storage_capacity Total storage capacity available.
    storage_overhead Overhead for management services and storage redundancy.

Example

Table with example VM specification:
VM specification Description
50 VMs of Type 1 Each VM requires 2 vCPUs, 8 GiB of memory, and 100 GiB of storage.
20 VMs of Type 2 Each VM requires 4 vCPUs, 16 GiB of memory, and 200 GiB of storage.
Table with example infrastructure Resources (Before Overhead and HA)
Infrastructure Resources (Before Overhead and HA) Description
Compute Nodes 13 nodes, each with 64 cores and 2 TB (2048 GiB) of memory.
Total Raw Storage 16 nodes (13 compute + 3 control), each with 10 disks of 7.68 TiB each.
Table with example overhead and high availability reserves
Overhead and high availability reserves  
CPU: Overhead 32 cores reserved for management services.
CPU: High Availability (HA) Reserve 20% of the remaining vCPUs.
CPU: Control Plane Requirements 4 cores are reserved for control plane operations.
Memory: Overhead 72 GiB reserved for management services.
Memory: High Availability (HA) Reserve 20% of the remaining memory.
Memory: Control Plane Requirements 24 GiB reserved for control plane operations.
Storage: Total Raw Storage 16×10×7.68 TiB = 1228.8 TiB
Storage: Usable Storage After 3-Way Replication (1228.8)/3 TiB​ = 409.6 TiB
Storage: Storage Overhead Assume 10% overhead for management services and system needs: 40.96 TiB
Storage: Total Available Storage After Overhead Total Available Storage= (Total Usable Storage) - (Storage Overhead)= 409.6 TiB - 40.96 TiB= 368.64 TiB
Table showing CPU calculation
CPU Value
Total Physical CPU Cores Total Physical CPU Cores= 64×13= 832 cores
Total vCPUs 128×13= 1664 vCPUs (assuming each core provide 2 vCPUs , per node would be 64×2 =128)
Control Plane Requirements 4 cores are reserved for control plane operations.
Overhead 32 cores reserved for management services.
High Availability (HA) Reserve 20% of the remaining vCPUs.
Total Available vCPUs After Control Plane, Overhead, and HA Reserve Total Available (vCPUs) = (vCPUs_per_node × node_count − control_plane_requirements − overhead) × (1−ha_reserve_percent) × (overcommit_ratio)
Substituting the values Total Available vCPUs=(1664 - 4 - 159.63) × 0.8 × 1 = 1200.296 vCPUs (Overcommit ratio is assumed to be 1)

Rounding down, we have 1200 vCPUs available for VM deployments.
Comparison with Available CPU 180 vCPUs required < 1200 vCPUs available
Result: The infrastructure has sufficient CPU resources to support the required virtual machines even after considering the control plane requirements, overhead, and HA.
Table showing memory calculation
Memory Values
Total Physical Memory Total Physical Memory=2048×13=26624 GiB
Control Plane Requirements 24 GiB reserved for control plane operations.
Overhead 72 GiB reserved for management services.
High Availability (HA) Reserve 20% of the remaining memory.
Total Available Memory After Control Plane, Overhead, and HA Reserve Total Available Memory (GiB) = (total_node_memory × node_count − control_plane_requirements − overhead) × (1 − ha_reserve_percent)
Substituting the values Total Available Memory (GiB)=(2048×13 − 24 − 72) × 0.8= 26528×0.8= 21222.4 GiB

Rounding down, we have 21222 GiB available for VM deployments.
Comparison with Available Memory 720 GiB required < 21222 GiB available

Result: The infrastructure has sufficient memory resources to support the required VMs even after considering the control plane requirements, overhead, and HA.

Table showing storage calculation
Storage Values
Total Raw Storage Total Raw Storage=16×10×7.68 TiB= 1228.8 TiB
Total Usable Storage After 3-Way Replication Total Usable Storage=(1228.8)/3 TiB= 409.6 TiB
Storage Overhead Storage Overhead= 40.96 TiB
Total Available Storage After Overhead: Total Available Storage=409.6 TiB − 40.96 TiB= 368.64 TiB
Total Storage required for Type 1 VMs: Storage for Type 1 VMs= 50×100 GiB= 5000 GiB= 5 TiB
Total Storage required for Type 2 VMs: Storage for Type 2 VMs= 20×200 GiB= 4000 GiB= 4 TiB
Total Storage required for all VMs: Total Storage required= 5 TiB+4 TiB= 9 TiB
Comparison with Available Storage: 9 TiB required < 368.64 TiB available
Conclusion
  • CPU: 1200 vCPUs available; 180 vCPUs required (Sufficient)
  • Memory: 21222 GiB available; 720 GiB required (Sufficient)
  • Storage: 368.64 TiB available; 9 TiB required (Sufficient)

The infrastructure can support the deployment of 50 Type 1 VMs and 20 Type 2 VMs, with sufficient CPU, memory, and storage resources. These resources are available after accounting for overhead and high availability reserves.