Support for Object Placement in Wide Area Heterogeneous Distributed Systems

One of the open challenges in distributed computing systems is determining how to place tasks onto processors when they are needed (in the Legion project being developed at UVA the basic computational units are modelled as objects, so the problem is one of object placement). The placement decision is crucial because it determines the run-time behavior of an object, including performance, cost and whether it can run at all. Many approaches have been developed to address this problem in a distributed system environment, but it is our claim that these efforts do not take the proper approach for supporting the needs of the large wide area heterogeneous virtual computer systems we envision will exist in the future. In particular, the systems developed to date are inadequate because they 1) focus on solutions for a narrow set of application types, environments, or user objectives, and 2) often inadequately support the full complexity and features of large distributed systems. We propose to better support the placement process in distributed systems by employing a new approach. Our approach is different from previous ones in that we propose to design a framework for supporting a wide range of different placement problems, user objectives and placement algorithms, rather than building a system that supports a single placement technique. The goal of the framework is to provide programmers with the basic mechanisms to support each facet of the placement process which will enable them to implement the placement policies and techniques that meet their needs. On the other hand, individuals will be competing for limited resources owned by different people or organizations. Therefore, the framework must also contain mechanisms to enforce the policies of resource owners and to resolve conflicts between users. The research effort proposed here will focus on developing the mechanisms needed to support flexible distributed object placement. To identify the main components of the placement process and the key issues that must be resolved, we will first develop a general model of the placement process. Using this model, we will next develop and implement a framework to support object placement within the Legion system. Developing such a framework will lend insight into our placement model and will also provide a proof of concept for our approach. Finally, we will demonstrate the usefulness of our framework approach by mapping a range of placement algorithms to the Legion framework and evaluating the performance of several algorithms versus that provided by the system today.