VMware vCloud Director for Service Providers creates an abstraction layer on top of vSphere that pools data center resources, such as compute, memory, networks, and storage, along with their associated policies, into Virtual Data Centers (VDCs). In a vCloud Director environment, all physical hardware resources are presented first to vCenter Server, and are subsequently carved up into logical data centers, referred to as Provider Virtual Data Centers (PVDCs). The consumer's organization, defined within vCloud Director, draws on the PVDC resources through the pooled resources assigned to their respective Organization Virtual Data Centers (Org VDCs). Each Organization VDC can only consume resources from a single Provider Virtual Data Center, although a PVDC can provide resources to multiple Organization VDCs. vApps represent the consumer's organization deployed workloads, which reside within the tenants’ Org VDC. This architecture is illustrated in the following figure.

 

 

By employing these constructs, vCloud Director allows VMware Cloud Providers to create logical data centers, composed of compute, network, and storage resources, mapped from the underlining physical components. These logical data centers are referred to as Provider VDCs, and play an important part in employing vSAN as part of a vCloud Director public cloud platform. However, in addition to this, Provider VDCs pose the following design considerations:

As outlined previously, the vCloud Director consumer's vApps represent the tenants’ deployed workloads, the resource requirements for which can vary significantly. As with all applications, some vApps require more compute resources, while others might require more disk-intensive operations to be processed. vCloud Director, by design, allows Organization VDCs to be categorized based on workload requirements, or tier of performance required.

To address these varying workload requirements, Provider VDCs must typically be configured to offer multiple service levels within the vCloud environment. These are sometimes categorized as either precious metals, such as Gold, Silver and Bronze, or as various tiers of service, which may include Tier 0, Tier 1, Tier 2, or Tier 3.

However, many service providers also avoid these types of labels. After all, vCloud consumers might not like the idea of placing workloads on lower tiers or lower values of storage. Therefore, this document categorizes storage as either Fast, Standard, or Capacity. The following table provides an example of tiered Provider VDCs.

Accurately capturing tenants’ application performance requirements, which must be satisfied by each level of service, remains a challenge for VMware Cloud Providers. While guest operating systems typically have well-understood requirements, it is more difficult to quantify application performance requirements, particularly when it comes to storage. In addition, workload storage requirements can be measured in various ways. Some vCloud consumers define their requirements into read and write IOPS, maximum latency requirements, and sequential or random I/O profiles, whereas other consumers of services define their storage requirements from an application perspective, and request resources to meet specific application requirements.

Either way, capturing tenants’ workload performance requirements can be challenging for VMware Cloud Providers who might have little or no control over storage tiering and pricing of the VMs being deployed. While applications and operating systems typically have well-defined requirements for CPU and memory consumption, it is often far more difficult to quantify the performance characteristics of storage resources. For instance, a customer might simply request storage to support the capabilities outlined in the following table.

It is important to define both performance and capacity requirements from vCloud consumers, because often vApp storage profiles are not simply about performance or capacity, but are a balance between the two, which can be the storage design challenge within a vCloud Director environment.

Providing storage resources that can meet all of the various performance requirements set out by tenants, while also providing the ability to scale and manage changes or spikes in demand for resources, are key design requirements for VMware Cloud Providers when implementing vCloud Director for Service Providers with vSAN storage resources. The key aim for any service provider is to allocate storage as efficiently as possible across the platform, so that resources can be utilized by Organization VDCs with a minimum of waste and optimum performance, delivering operational stability and accurate billing. This optimized configuration and allocation of storage across Provider VDCs and consumption by Organization VDCs is vital to achieving an efficient tiered storage model.