A federated time distribution system for online laboratories

Abstract

The iLab project began in June of 2000 with the initial goal of creating Internet accessible laboratory experiments. After the successful implementation of several distinct laboratories, the goals of the project shifted to address the design and construction of a generalized infrastructure capable of supporting a wide variety of laboratory experiments. Early experiences suggested the need for a configurable experiment scheduling system. Such a system would be particularly important in the face of expected growth: as the number of providers and consumers grew, it would become crucial to empower providers with the ability to enforce experiment usage policies, to guarantee timely lab access to clients, and to maximize resource usage whenever possible. We will explore how the present iLab infrastructure can be modified to allow for experiment scheduling. This system would be designed in keeping with two key principles: generality and architectural consistency. It would have to support disparate scheduling algorithms of varying complexity and remain faithful to the theme and priorities of the existing iLab infrastructure. Design will be based on requirements gathering and the analysis of existing remotely available experiments. Resulting changes to the iLab infrastructure will be enumerated, justified, and their ramifications discussed. This design will be implemented and considered in the same fashion. Finally, future scheduling work within the context of iLab will be described.