Assuring end-to-end timeliness properties in a distributed real-time system poses a variety of challenges at each, of several, levels of abstraction. The challenges vary with levels of abstraction that correspond generally to levels of architectural scale. In particular, there are key differences in the design forces for resource scheduling (1) within the operating system kernel, (2) within the middleware infras-tructure on a single endsystem, and (3) distributed across endsystems in the entire system. While there is signifi-cant commonality related to the system’s end-to-end time-liness requirements, the forces and their resolution vary non-trivially with each level of abstraction. Some design forces are seen at, or even span, multiple levels of scale. We document patterns that offer an organizing approach to design, within which the design forces at and across each level of abstraction can be resolved. This paper describes a pattern language for rationalizing resource scheduling at multiple levels of scale in distributed real-time and embedded systems.