S. Chatterjee. Dynamic application structuring on heterogeneous, distributed systems. In Proc. IPPS/SPDP'99 Workshop on Parallel and Distributed Real-Time Systems, 1999.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....at the level of individual components (e.g. the network congestion induced dynamic resizing of the acknowledgment window in the TCP protocol) little support is available for structuring general purpose resource aware applications. Several researchers have begun to address this shortcoming [9, 15, 16, 19, 21]; however, most such efforts place a substantial burden on application developers requiring them to provide explicit specification of both resource utilization profiles (which resources are used at which time and in what quantity) and adaptation behaviors (how should the application react to ....

....to changes in the environment. The ActiveHarmony [15] and AppLeS [3, 22] projects provide application level mechanisms and resource monitoring tools to enable general applications to adapt to changing resource characteristics, albeit with a fair degree of user involvement. EPIQ [21] and ERDoS [9] projects are closer to our approach in that they expose quality aspects of an application, automatically trading off output quality against resource requirements. However, they focus on the negotiation aspects of configuration and an analytic specification of the benefit utility function. Our ....

S. Chatterjee. Dynamic application structuring on heterogeneous, distributed systems. In Proc. IPPS/SPDP'99 Workshop on Parallel and Distributed Real-Time Systems, 1999.

....of application tunability; however, unlike the focus of this paper, DRMS exploits the reconfigurability primarily to optimize system utilization while still providing only best effort guarantees to the application. Our work is probably most closely related to a few recently started projects [23, 18, 5, 26, 20, 15, 9] that are looking into the problem of adapting application behavior in response APPLICATION TUNABILITY IN PARALLEL AND DISTRIBUTED SYSTEMS 21 to variations in system resources. Despite the shared goals, our approach differs from these projects in several important ways: The Darwin project [23] ....

....entities are mapped differently but to the same kind of physical resources. Our notion of tunability is much more general, enabling the application to trade off its resource requirements over several dimensions including time, output quality, and resource type. The EPIQ [20, 15] and ErDos [9] projects are closer to our approach in that they look into the quality aspects of an application, trading off output quality against resource requirements. Both EPIQ and ErDos take the view that the quality of an execution degrades when there are insufficient resources. Our work stresses the ....

S. Chatterjee. Dynamic application structuring on heterogeneous, distributed systems. In Proc. IPPS/SPDP'99 Workshop on Parallel and Distributed Real-Time Systems, 1999.

....at the level of individual components (e.g. the network congestion induced dynamic resizing of the acknowledgment window in the TCP protocol) little support is available for structuring general purpose resource aware applications. Several researchers have begun to address this shortcoming [9,19,20,23,25]; however, most such efforts place a substantial burden on application developers requiring them to provide explicit specification of both resourceutilization profiles (which resources are used at which time and in what quantity) and adaptation behaviors (how should the application react to ....

....adapt to changes in the environment. The ActiveHarmony [19] and AppLeS [3,26] projects provide application level mechanisms and resource monitoring tools to enable general applications to adapt to changing resource characteristics, albeit with a fair degree of user involvement. EPIQ [25] and ERDoS [9] projects are closer to our approach in that they expose quality aspects of an application, automatically trading off output quality against resource requirements. However, they focus on the negotiation aspects of configuration and an analytic specification of the benefit utility function. Our ....

S. Chatterjee. Dynamic application structuring on heterogeneous, distributed systems. In Proc. IPPS/SPDP'99 Workshop on Parallel and Distributed Real-Time Systems, 1999.

....of application tunability; however, unlike the focus of this paper, DRMS exploits the reconfigurability primarily to optimize system utilization while still providing only best effort guarantees to the application. Our work is probably most closely related to a few recently started projects [5, 9, 15, 18, 20, 23, 26] that are looking into the problem of adapting application behavior in response to variations in system resources. Despite the shared goals, our approach differs from these projects in several important ways: The Darwin project [23] permits the application to specify its resource requests in the ....

....entities are mapped differently but to the same kind of physical resources. Our notion of tunability is much more general, enabling the application to trade off its resource requirements over several dimensions including time, output quality, and resource type. The EPIQ [15, 20] and ErDos [9] projects are closer to our approach in that they look into the quality aspects of an application, trade off output quality against resource requirements. Both EPIQ and ErDos take the view that the quality of an execution degrades when there are insufficient resources. Our work stresses the ....

S. Chatterjee, Dynamic application structuring on heterogeneous distributed systems, in Proc. IPPS#SPDP '99 Workshop on Parallel and Distributed Real-Time Systems," 1999.

....for adaptation at the level of individual components (e.g. the network congestion induced dynamic resizing of the acknowledgment window in the TCP protocol) little support is available for structuring such globally adaptive applications. Several researchers have begun to address this shortcoming [10, 14, 15, 17, 18]; however, most such efforts place an unreasonable burden on application developers requiring them to provide explicit specification of both resourceutilization profiles (which resources are used at which time and in what quantity) and adaptation behaviors (how should the application react to ....

....variations would require better algorithms that take into consideration the sensitivity of application configurations to resource variations so as to not degrade overall performance by unnecessary adaptations. 8 Related Work Our work is most closely related to a few recently started projects [3, 10, 14, 17, 18, 19] that are looking into the problem of adapting application behavior in response to variations of system resources. The Darwin project [17] permits an application to specify its resource requests in the form of a virtual mesh of nodes (representing desired services) and edges (denoting ....

[Article contains additional citation context not shown here]

S. Chatterjee. Dynamic application structuring on heterogeneous, distributed systems. In Proc. IPPS/SPDP'99 Workshop on Parallel and Distributed Real-Time Systems, 1999.

No context found.

S. Chatterjee. Dynamic application structuring on heterogeneous, distributed systems. In Proc. IPPS/SPDP'99 Workshop on Parallel and Distributed Real-Time Systems, 1999.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC