25
Taking into these considerations for peaks and different types of app/desktop
virtualization architectures one must select and design a storage solution that is
capable of meeting the peak boot, login, and steady state demands of the environment.
To understand the storage requirements of the design, one should perform a desktop
assessment on the existing physical PC environment. This desktop assessment will
gather the real performance and capacity details from the user base so that one can
apply these to the design calculations.
A final thought on app/desktop virtualization-related storage requirements is that
aside from being very unpredictable in the I/O side of things, desktop workloads are
also very write heavy. Unlike many server workloads that are mostly reading data
and serving it to users, desktops are typically spending more time writing to disk.
Writes are more intensive to the storage array than reads are. A typical server workload
might be 80% reads and 20% writes, while the steady state virtual desktop workload
might be the opposite. When evaluating your storage choices, be sure to pay close
attention to how the storage solution buffers and commits writes, versus some large
promise that it does an "excellent job" at caching commonly read blocks.