3.1.1 Service Design Parameters
As part of the design process for the Virtual Private Cloud on-demand service offering, give special consideration to key service settings and values in vCloud Director that can impact service performance and consistency levels for a subscriber’s Organization virtual data center. Given the pay-as-you-go allocation model employed in this service, certain circumstances might arise that result in subscribers overcommitting resources over time. If not properly managed, these circumstances can negatively affect performance for all application workloads. The following table provides an example of these key service settings, values, and justifications.
Table 10. Resource Allocation Settings Example – VPC On-Demand Service Offering
Resource Type | Value Range | Sample Setting | Justification |
CPU allocation | Variable (GHz) based on physical host capacity | 50 GHz | Maximum amount of CPU available to the virtual machines running in the target organization’s virtual data center (taken from the supporting provider virtual data center). |
CPU resources guaranteed | 0–100% | 0% | Percentage of CPU resources that are guaranteed to a virtual machine running within the target organization virtual data center. This option controls over-commitment of CPU resources. |
vCPU speed | 0–8 GHz | 1 GHz | Value defines what a virtual machine or vApp with one vCPU consumes at most when running within the target organization virtual data center. A virtual machine with two vCPUs consumes a maximum of twice this value. |
Memory resources guaranteed | 0–100% | 75% | Percentage of memory that is guaranteed to a virtual machine running within the target virtual data center. This option controls over-commitment of memory resources. |
Maximum number of virtual machines | 1–Unlimited | Unlimited | Safeguard that allows control over the total number of vApps or virtual machines a subscriber can create within the target virtual data center. |
In this example, the minimum vCPU speed setting is configured at 1 GHz (1000 MHz), with a memory resource guarantee of 75 percent. CPU resource guarantees and limitations on the maximum number of virtual machines supported per tenant are optional and can be implemented at the provider’s discretion. The provider can use the combination of these settings to change over-commitment from aggressive levels (for example, resource guarantees set to less than100 percent) to more conservative levels (for example, resource guarantees always set to 100 percent), depending on SLAs in place or fluctuating service loads.