Real-Time Design with Peer Tasks

We introduce a real-time design methodology for large scale, distributed, parallel architecture, real-time systems (LDPARTS) that approaches system scheduling analysis in a way different from those methods that use a scalar metric of urgency such as found in rate (or deadline) monotonic theories. The latter assume the place for scheduling prioritization to be at the functional level of run-time processes. For example, in the Ada programming language this refers to task scheduling. In our method, the fundamental units of prioritization, which we call work items, are system level or domain specific objects with timing requirements (deadlines) associated with them in the requirements specification. For LDPARTS, a work item consists of a collection of tasks. No priorities are assigned to tasks or, equivalently, tasks have equal priorities. Such a collection of tasks is referred to as peer tasks. Current scheduling theories are applied with artifact deadlines introduced by the designer whereas our method schedules work items to meet specification deadlines (sometimes called end-to-end deadlines) required by the user. The new method supports these scheduling properties. The scheduling of work items is based on domain specific importance rather than task level urgency and still meets as many work item deadlines as can be met by scheduling tasks with respect to urgency. Second, the minimum (closest) on-line deadline that can be guaranteed for a work item of highest importance, scheduled at run time, is approximately the inverse of the throughput, measured in work items per second. Third, throughput is not degraded during overload and instead of resorting to task shedding during overload, the designer can specify which work items to shed. We prove these properties in a mathematical model (ref. 1). The degree to which these hold for a specific system depends on how small the statistical variance of work item throughput is over the required system performance envelope. The method works best in a project with a " cut and try " iterative development approach, where measurement of work item throughput may be made under as realistic system load conditions as possible.