Performance: Even a properly scaled infrastructure will experience performance issues due to usage spikes, network lag, and underperforming or less-than-optimized servers. All of these can be managed to some degree; usage spikes can be handled with oversizing the hardware; underperforming or lessthan-optimized servers and networks can be actively monitored and maintained. While this can improve performance, it also requires additional investment in hardware for usage spikes and additional and skilled IT resources for monitoring and tuning the servers and network. Many network lag issues are the result of third-party services and cannot be avoided even by the bestrun IT organizations using top-of-the line equipment.
System/network failure: Although virtualization can provide a more stable environment than intelligent desktops, it can also magnify server or network failures. The cost of ensuring high availability and system and network redundancy must be included in the TCO analysis. Organizations must also balance the risks of system and network failure and the costs of mitigating that risk against the cost savings of virtualization.
16 Intel IT Center Planning Guide | Desktop Virtualization