When a Cloud Provider has two data centers within close proximity to each other, it is possible to distribute the elements of the management environment across a single vCenter Server controlled resource platform which is itself distributed across the two data centers creating a single, logical data center. The following figure shows the vCloud Director instance deployed within such a management resource environment. The possible workload vCenter Server and resource cluster topologies are the subject of the following sections of this document, so are intentionally omitted.

 

 

This figure shows the Management vCenter which controls resources in data center “A” and “B”. The two data centers are interconnected with sufficient bandwidth and low enough latency to enable a single compute cluster and its associated storage to operate across both. Typically, the elements of the vCloud Director instance run in one of the two data centers, and fails over to the other in the event of a failure within the active site. Some of the failover can be achieved using VMware vSphere High Availability. Other more complex elements (such as integration into external systems) might require a recovery automation tool such as VMware Site Recovery Manager™.

While it is, therefore, possible to stretch the vCloud Director management components between sites, the details of implementing a stretched management environment are out of scope for this document. Within the remainder of the document, “stretched” refers to the workload resources controlled by a single vCloud Director instance being deployed across multiple locations rather than the specific topology of the management components of vCloud Director itself. In subsequent figures, because the management resource topology is omitted, the vCloud Director instance is shown as the outermost container within, or across, data centers.