2.3.1 Dual vCenter Server Instances – Stretched vCloud Director
In the first model, vCloud Director is installed in a single location, but manages resources under the control of VMware vCenter Server instances in each site. The maximum round-trip network latency from vCloud Director to vCenter Server, which had previously been 20ms, was increased to 40ms with v8.20 allowing one vCloud Director instance to manage workload vCenter Server instances in multiple locations. The following figure shows a representation of the relationships between the elements in this model. Tenant 1 has separate resources within Provider VDC “A” and “B”. Those resources, while separate at the compute/storage layers, are connected at the network layer either by a common or shared vCloud Director network or through data center interconnects outside of vCloud Director.
Figure 4. Dual vCenter Server Stretched vCloud Director Object Hierarchy
When using this model, Cloud Providers must rely on vCloud Director deployed on one of the sites, with a disaster recovery (DR) plan to recover the management platform to another site in the event of a failure. This DR plan mitigates against a complete site failure at the “primary” vCloud Director site leaving the other operational, but without a functioning customer or provider management portal.
In the event of a failure at the “secondary” or “remote” site, the remaining site operates as a single-site vCloud Director deployment with a single, local vCenter Server as shown in
Figure 3. However, should the failure affect the primary site, while the remaining vCenter Server and its associated workload hosts remain unaffected at the secondary site, initially at least, there is no vCloud Director instance from which to control them. Recovering the vCloud Director to the remaining site is represented in the following figure.
Figure 5. Dual vCenter Server Stretched vCloud Director in Disaster Recovery
Recovery of the vCloud Director instance complete with vCloud Director cells, databases, message bus, and other supporting infrastructure, is not a trivial task. While VMware Site Recovery Manager can be used to automate much of this process, additional steps might be necessary to affect the network topology changes for the remaining vCenter Server and other Business or Operational Support Systems (BSS/OSS) to restore connectivity to the vCloud Director instance in its new location.
