Customers deploying Branch Repeater for providing WAN acceleration and optimizations to Citrix XenApp and XenDesktop solutions or for optimizing any other TCP based protocols have a choice to either deploy the physical appliances or the virtual appliances (VPX). While deploying physical appliances has been a well-known territory for many years, especially in terms of its specifications and expected scalability, deploying VPX instances warrant answering a few important questions around performance and scalability, such as:

  1. What are the recommended resource assignments for VPX instances? Are they workload specific or the same regardless of variations in workloads?
  2. Do the optimizations provided by VPX depend on how well its components (vCPU, Disk and Memory) can scale?  Will the performance benefits be impacted if VPX scalability limits are reached?
  3. Is it recommended to share VPX instances with other VMs on the same hypervisor host? How does this sharing of VMs impact VPX and the host hypervisor scalability?
  4. What are the recommendations for scaling-out VPX instances by load balancing them with Citrix NetScaler? How does load balancing multiple VPXs affect its scalability? Does a load balanced VPX solution scale-out linearly?

Within Worldwide Consulting Solutions, we conducted  thorough VPX scalability analysis where we tested Branch Repeater VPX resources such as vCPU, memory and disk I/O under high bandwidth usage scenarios with simulated XenDesktop HDX™ Mediastream traffic.  The Branch Repeater VPX was evaluated with up to 45 Mbps of traffic (the licensed limit for a single VPX) routed through the virtual appliance. While it is expected that the scalability factors will differ to an extent based on variations in workloads and other environmental factors, this testing allows us to provide some important guidance around VPX scalability and monitoring recommendations for resources optimizing XenDesktop workloads. And most importantly to allow you in identifying the scalability needs for your environment and avoid any impact in achieving maximum performance.

Here is the detailed report for your read –

I find your feedback to be valuable so please feel free to reach out with questions and/or comments.


Bhumik Patel