Leveraging vSAN for Highly Available Management Clusters : Cloud Service Provider Use Case : 3.2 Cloud Service Provider Considerations : 3.2.1 Multisite Service Provider Considerations
   
3.2.1 Multisite Service Provider Considerations
Locate the management cluster and the related resource groups at the same physical site to provide a consistent experience in the level of service provided, because some of the management components might have restrictions when deployed across geographically distributed networks (for example, vCloud Director support for multisite deployments). The main reasons to recommend same-site management is to optimize the connectivity and reliability by avoiding possible latency issues. Nevertheless, the management across distributed topologies or stretched clusters across multiple sites are feasible solutions that can be assessed as viable on a case-by-case basis.
A Cloud Service Provider might operate a more complex topology of data centers with multisite scenarios to manage and supervise. Some of these scenarios might include, but are not limited to:
MAN and/or WAN topologies with their connectivity speed and latency.
Network layer available between sites (Layer 2 or Layer 3).
Resource clusters offering services for workloads that require either stretched or separated configurations across the multisite topologies.
The options and advantages to deploy localized management clusters in each site are related to each characteristic workload:
Multisite topologies with WAN or MAN connectivity having an RTT over 20 ms are not suited for deployment of single-site style management infrastructures including vCloud Director.
Multisite topologies with WAN or MAN connectivity having RTT below 20 ms, but with Layer 3 only connectivity, still require separate management clusters.
Multisite topologies with WAN or MAN connectivity having RTT below 20 ms and Layer 2 connectivity offer the choice between stretched or separate management clusters.
In the event of a site failure and subsequent recovery, the management functions are the first and most rapid to be recovered and so are preferred to be at the same location of the resources to be managed.
vSAN recommends communications across data nodes to be over Layer 2, and for witness only communications to occur over Layer 3.
vSAN and vSphere FT are not supported together in a stretched virtual SAN cluster deployment.