Supporting Deadline Constrained Distributed Computations on Grids

Zhao, Xinghui; Jamali, Nadeem

doi:10.1109/grid.2011.29

Cited by 9 publications

(9 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, although the computational overhead of reasoning in the system is far below the benefit of doing it, we want to explore opportunities for explicitly balancing the overhead involved in reasoning against the quality of the schedule required. We hope to build on our previous work implementing a tuner facility for balancing the computational cost of creating fine-grained processor schedules against the cost of carrying out the actual computations [21]. The tuner carries out meta-level resource balancing between the reason and the computations being reasoned about; its parameters can be set manually or be set to self-tune at run-time in response to observations about the ongoing computation.…”

Section: Resultsmentioning

confidence: 99%

“…In our previous work, we have constructed DREAM a (Distributed Resource Estimation and Allocation Model) [20] and related mechanisms [21] for reasoning about scheduling of deadline constrained concurrent computations over parallel and distributed execution environments. In the most recent work [22], this approach have been repurposed to achieve dynamic load balancing for computations which do not constrained by deadlines.…”

Section: Reasoning About Multicore Energy Consumptionmentioning

confidence: 99%

See 1 more Smart Citation

Energy-aware resource allocation for multicores with per-core frequency scaling

Zhao

Jamali

2014

J Internet Serv Appl

Self Cite

View full text Add to dashboard Cite

With the growing ubiquity of computer systems, the energy consumption of these systems is of increasing concern. Multicore architectures offer a potential opportunity for energy conservation by allowing cores to operate at lower frequencies when the processor demand low. Until recently, this has meant operating all cores at the same frequency, and research on analyzing power consumption of multicores has assumed that all cores run at the same frequency. However, emerging technologies such as fast voltage scaling and Turbo Boost promise to allow cores on a chip to operate at different frequencies. This paper presents an energy-aware resource management model, DREAM-MCP, which provides a flexible way to analyze energy consumption of multicores operating at non-uniform frequencies. This information can then be used to generate a fine-grained energy-efficient schedule for execution of the computations -as well as a schedule of frequency changes on a per-core basis -while satisfying performance requirements of computations. To evaluate our approach, we have carried out two case studies, one involving a problem with static workload (Gravitational N-Body Problem), and another involving a problem with dynamic workload (Adaptive Quadrature). Experimental results show that for both problems, the energy savings achieved using this approach far outweigh the energy consumed in the reasoning required for generating the schedules.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Reasoning About Multicore Energy Consumptionmentioning

confidence: 99%

Energy-aware resource allocation for multicores with per-core frequency scaling

Zhao

Jamali

2014

J Internet Serv Appl

Self Cite

View full text Add to dashboard Cite

show abstract

“…In our previous work, we have shown that this approach can be effective in achieving timeliness goals of computations [35]; here, we show that a similar approach -prioritizing message deliveries based on bounded knowledge of the computations they would lead to -can also improve overall performance of the system even in the absence of timeliness requirements. Particularly, we attempt to expedite delivery of the messages which would lead to computations which in turn would send messages of their own; these messages are potential bottlenecks.…”

Section: Fine-grained Resource Managementmentioning

confidence: 86%

Load balancing non-uniform parallel computations

Zhao

Jamali

2013

Proceedings of the 2013 Workshop on Programming Based on Actors, Agents, and Decentralized Control

Self Cite

View full text Add to dashboard Cite

Dynamic load balancing is critical in achieving high performance in parallel systems, especially for applications with unpredictable workloads. Traditional load balancing approaches such as work-sharing and work-stealing often assume that the computations can be divided into smaller computations based on some granularity. These approaches do not scale when the computations are not dividable and their sizes are highly variant. In this paper, we present a novel approach -founded in the Actor model known for its programmability -for dynamically load-balancing nonuniform parallel computations. Specifically, our approach is to explicitly reason about resource requirements and commitments at run-time in order to make fine-grained scheduling decisions. We prototype our approach by extending an efficient Java implementation of Actors, ActorFoundry. We have also implemented a tuner which dynamically balances the resources used by the computations vs. those used by the reasoning mechanism, essentially balancing the quality of scheduling against the resources needed for achieving it. We propose a new benchmark -the Unbalanced Cobwebbed Tree (UCT) -to fairly compare our approach to existing approaches, which captures the non-uniform nature of computations. Experimental results show that despite the higher overhead of providing Actors' programmability features, for a diverse enough set of computations, our approach outperforms both work-sharing and work-stealing approaches, and shows better scalability.

show abstract

“…Note that this function preserves the form of MWM+ (5) with the modification motivated by optimal weights in (6). Indeed, the fraction 1 1+dij(t)−t is similar to p ij (t) since 1/p ij (t) is the expected deadline of a packet in the formulation of the previous section.…”

Section: A Weight Functionmentioning

confidence: 93%

“…The applicability of deadline aware scheduling algorithms reaches as far as grid computation, in which threads much be adaptively scheduled to optimize distributed computation [6].…”

Section: Introductionmentioning

confidence: 99%

Deadline aware packet scheduling in switches for multimedia streaming applications

Bommannavar

Apostolopoulos

Bambos

2013

2013 IEEE International Conference on Communications (ICC)

View full text Add to dashboard Cite

We consider the problem of scheduling packets in an input queued switch with a focus on processing streamed multimedia data. In such applications, packets arrive with hard service deadlines; after the deadline for a packet has passed, it is no longer useful and does not get delivered -it is dropped. We seek policies to minimize the number of late packets, which are then dropped. The problem is formulated in a Dynamic Programming framework and shown to be intractable. The formulation is contrasted to the related crossbar switch scheduling problem, with an emphasis on the fact that we have a different objective function. A simplified probabilistic version of the streaming problem is used as motivation for a heuristic solution. Finally, we present results from a simulation in which a simple heuristic based on weighting queues according to the deadline of the leading packet consistently outperforms the wellknown maximum weight matching (MWM) algorithm.Index Terms-scheduling; deadlines; crossbar switch; inputqueued switch; maximum weight matching I. INTRODUCTIONIn this paper we consider the problem of moving data packets from a set of input ports to a set of output ports with the special provision that packets carry a deadline for delivery. Such a constraint appears in applications where the data are processed in a real-time streaming fashion; instead of incurring additional delay beyond a specified deadline, it is preferable to drop the late packet and move on to the next one. This appears frequently in the context of Voice over IP (VoIP), Video over Broadband (VoBB), Video on Demand (VoD) and a host of other applications in which delay beyond a specified threshold is no longer acceptable.Results for queuing problems involving packet deadlines appear frequently in the multimedia streaming and wireless communications literature. In [1], the problem of scheduling multiple queues in which streaming packets have deadlines over a wireless channel is considered, but there are not multiple output queues. The analysis in [2] examines policies that are simultaneously aware of channel strength, packet deadlines and distortion and [3] looks at scheduling packets at a base station to maximize quality of service to downlink users where deadlines play a role in the quality. Policies for multicast routing with similar loss criteria and packet constraints are studied in [4]. Packet deadlines as well as packet values are considered in [5] for a single queue, and bounds on the competitive ratio are derived.

show abstract

Supporting Deadline Constrained Distributed Computations on Grids

Cited by 9 publications

References 14 publications

Energy-aware resource allocation for multicores with per-core frequency scaling

Energy-aware resource allocation for multicores with per-core frequency scaling

Load balancing non-uniform parallel computations

Deadline aware packet scheduling in switches for multimedia streaming applications

Contact Info

Product

Resources

About