Will your Citrix XenApp infrastructure perform for thousands of mobile users?
The number of mobile devices in our companies are exploding – and one of the fastest and best ways to deliver all those existing Windows-based apps to these devices is using app virtualization. At the same time many of you are probably thinking about upgrading and expanding your Citrix XenApp deployment to Citrix XA 6.5 or XD 7 in any case.
But can your Citrix XenApp infrastructure support this upgrade and all these new users? When customers started asking us these questions we decided we needed to check it out. So we asked Principled Technologies to look at helping us test out how well our Cisco UCS architecture with Cisco UCS VM-FEX would cope with these stresses.
- With Cisco UCS VM-FEX 29% more responsive Citrix XenApp sessions can be obtained
- Under extreme network loads, Cisco UCS VM-FEX enabled Virtual Machine (VM), provided up to 53% more CPU resource optimization, than is possible with a traditional vSwitch
In a server based computing (SBC) application such as Citrix XenApp hosted shared desktop, session responsiveness is key to creating the best possible end-end user experience. Under demanding network conditions, there can be challenges, to maintaining SBC session responsiveness and Cisco UCS VM-FEX offers the best solution to overcoming these challenges. Essentially, the Principled Technologies study demonstrated the power of Cisco UCS VM-FEX PCI-E pass-through in offloading network switching overheads to the Cisco Virtual interface card (VIC 1240 in this setup), reducing network latency and compute requirements, thereby resulting in a more responsive application. Refer to diagram below for details.
Test Bed setup
The Testbed was based on Login VSI 3.7 medium workload, and comprised two identical XenApp farms, one farm VM-FEX enabled while the other running a traditional vSwitch.
Better session responsiveness
As the number of user sessions was gradually increased, the user response times for both the XenApp farms were compared. As the number of user sessions increased to 140, Cisco UCS VM-FEX showed a noticeable improvement in session responsiveness. Figure-1 shows the enhanced session responsiveness with increasing sessions, of a VM-FEX enabled vSwitch when compared to a traditional vSwitch
Better CPU resource utilization under extreme network loads
This Test demonstrates the benefits of Cisco UCS VM-FEX switching overhead offload to the virtual interface card (VIC 1240). The graphs below show a consistent 53.1% reduction in CPU usage with Cisco UCS VM-FEX compared to a traditional vSwitch.
In addition, to the two quantified benefits described above, the Principled Technology study found that Cisco UCS VM-FEX provides simplified management and operations with the Cisco UCS Manager. CiscoUCS VM-FEX technology addresses the manageability and performance concerns by consolidating the virtual switch and physical switch into a single management point. The number of network management points is dramatically reduced, and physical and virtual network traffic both are treated in a consistent policy driven manner. Using Cisco UCS Manager, Cisco UCS VM-FEX configuration tasks such as VLAN access, MTU speeds were setup easily, without having to configure vSphere hosts individually. vSphere Hosts can be added to a UCS VM-FEX enabled switching envt easily, when compared to a manually intense operational environment presented by a traditional vSwitch
The switching method chosen for a server based computing envt, determines application performance and end-user experience. Switching to Cisco UCS VM-FEX provides a more responsive environment, allowing Citrix XenApp-UCS customers to enjoy the full benefits of virtualization, including dynamic workload management, without any compromises.
Read the Principled technologies whitepaper for details of the study