Performance of "Desktop in the Cloud" processing software deployment

Deploying desktop hydrographic software in the cloud as virtual PCs has been suggested as a bridging technology to fully cloud-aware processing solutions. We investigate the processing performance of such a system, examining different compute resources and storage options in a variety of operations, with an on-premises server as control. Our results demonstrate that all “desktop in the cloud” (DitC) deployments are slower than the control. A software requirement for a tightly integrated GPU can also significantly increase costs. These observations suggest that while feasible, this form of DitC is a sub-optimal model for implementing a cloud-based hydrographic data processing.