8. Multisite Considerations : 8.5 Supportability Considerations for Single Site Deployments
   
8.5 Supportability Considerations for Single Site Deployments
VMware supports vCloud Director 5.1 in MAN scenarios (as described in section 8.3, Multisite Terminology). Some supportability considerations are as follows:
*All provider workloads, with the exception of vCenter Server and vCloud Networking and Security Manager instances, must be deployed in a single location.
*Clusters backing provider virtual datacenters can be deployed in different locations if connectivity between locations has latency requirements as described in section 8.3, Multisite Terminology. Minimize the likelihood of a path failure across datacenters that might partition the provider workloads.
*vCenter Server and vCloud Networking and Security Manager instances managing and servicing clusters in distributed locations can be deployed either close to the vCloud Director core components (vCloud Director cells and vCloud Director database) or close to the clusters they manage.
Architects implementing a single site vCloud across different locations should deploy the various components with consideration given to sensitive operations such as vApp copies so that the deployment is fully optimized and the architecture takes into account network chokepoints (especially in terms of bandwidth) that can exist even in a MAN scenario. This has to do more with optimization than supportability.
Stretched clusters (which include stretched vSphere DRS clusters and stretched storage) are fully supported when implemented with the storage vendor-neutral guidance documented in this section. Stretched clusters (which require 10ms or lower latency) can enable increased flexibility in both tenant and provider workload placement.
NoteThe generic single site considerations in this section apply to tenant deployments that have latency within 20ms. Stretched clusters for tenant workloads are only supported when sites have latency within 5ms or 10ms (depending on the vSphere release and underlying storage technology used). At 5ms or 10ms latency (depending on the vSphere release and underlying storage technology used), the location of provider infrastructure components is more flexible. It is recommended that you:
Follow the above-recommended guidance for any single site deployment within 20ms.
Follow specific recommended guidance for vCloud Director deployments on top of vSphere stretched clusters within 10ms (these practices are dictated by the underlying storage solution supporting the stretched cluster, and may override the vendor neutral stretch cluster recommended practices in this document).
See the VMware vCloud Blog article, Stretched vCloud Director Infrastructure (http://blogs.vmware.com/vcloud/2013/01/stretched-vcloud-director-infrastructure.html) for more information.