DARPA ULTRALOG Final Report - Industrial and Manufacturing ...
DARPA ULTRALOG Final Report - Industrial and Manufacturing ...
DARPA ULTRALOG Final Report - Industrial and Manufacturing ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
architecture based on both the Grid <strong>and</strong> Web Service technologies. The Grid Service enables the<br />
integration of resources <strong>and</strong> services across distributed, heterogeneous, dynamic virtual<br />
organizations [3]. Cost <strong>and</strong> quality considerations may force large number of customers to look<br />
for resources <strong>and</strong> services via such an architecture to deal with their own computing problems.<br />
Ubiquitous computing technology embeds computers in various objects <strong>and</strong> places for sensing<br />
<strong>and</strong> controlling environments [4]. As this technology is becoming realized <strong>and</strong> gives rise to<br />
complex computing problems, the use of such an architecture might be inevitable.<br />
In a grid service environment, a problem is processed by composing multiple resources <strong>and</strong><br />
services. As there can be several alternative compositions of resources <strong>and</strong> services for a given<br />
problem, virtual markets will play a critical role in coordinating huge amount of economic<br />
entities such as customers, service providers, <strong>and</strong> resource providers. There are various market<br />
mechanisms such as OCEAN [5], Compute Power Market [6], <strong>and</strong> Nimord/G [7], proposed for<br />
the large-scale virtual economy. However, one essential enabler of such markets is the ability to<br />
agilely quantify the quality of service (QoS) achievable by each alternative. Without such a<br />
capability, the alternatives cannot be valuated in a timely manner <strong>and</strong> the virtual economy will<br />
fail to efficiently utilize the resources <strong>and</strong> services.<br />
There can be various ways of defining QoS depending on the nature of the problems. We<br />
consider a class of problems whose QoS is determined by completion time for generating a<br />
solution. The completion time (also called makespan) is one of the most widely studied<br />
objectives for diverse scheduling problems in the context of multiprocessor, manufacturing, <strong>and</strong><br />
project management. Regarding to problem solving structure we adopt component-based<br />
architecture as a general framework. A component is a reusable program element. Component<br />
technology utilizes the components so that developers can build systems needed by simply<br />
2