DescriptionComputational Science has rapidly grown over the past decades due to adoption of “commodity” technologies starting with Beowulf clusters of the 90’s, through building large computational resources based on commercially available servers. This trend led the scientific community to adopt commercial software practices such as virtualization and provisioning systems. As a result scientific applications are becoming increasingly less tied to specific hardware, increasingly diverse, and increasingly complex with many interconnected components and dependencies. Addressing problems requiring data/model synthesis or data sharing and analysis as services, considerations in this space typically take the form of managing heterogeneous sub-components, differing and at times conflicting dependencies, as well as diverse and changing deployment considerations. In this panel we will discuss the growing movement to address these challenges: changes in the type of support provided, conventions that would facilitate such activities, and tools that can be leveraged to simplify these types of applications.