In scenarios where companies have a Global presence with multiple geographical locations, it is sometimes desirable to have a Provisioning Services Farm that spans multiple locations across the globe.

PVS requires a constant and reliable connection to the SQL database. If there's a delay in communication between the PVS Server and the SQL database, this can cause performance and stability issues.
These issues might include, but are not limited to:

• Delays in the PVS Console that can cause the console to become unresponsive or crash  
• PVS Stream Process can experience hung threads causing it to self terminate. This in turn can cause PVS service outages  
• Target boot process can be severely affected to the point where the boot process can fail completely 

Citrix's leading practice when it comes to designing and deploying Provisioning Services in multi-geo environments is to ensure that PVS farms do not span data centers with a network latency that can affect communications between the PVS Servers and the SQL database. It's important to note that Network Latency is not the only variable to consider. Overall communications reliability history and performance must also be considered.

If having a global PVS Farm across data centers is required, Citrix recommends additional testing to confirm network performance and reliability, this is especially relevant for networks with higher latency (>15ms). If reliability and performance cannot be guaranteed then it is best to implement multiple PVS farms with local SQL databases for optimal performance and reliability.

It should also be noted that the majority of enterprise customers who have multiple geographical locations implement a "pod architecture", where each geographical location has its own PVS Farm and SQL database, regardless of whether the network connectivity across the locations could've sustained one global PVS Farm. This type of “pod architecture” maximizes performance and mitigates risk by reducing the size of failure domains.