2006/06/16 by Bin Bin Chen, Chen, Bin Bin, Pascale Vicat-Blanc Primet +2
Computer Science · #Distributed and Parallel Computing Systems #Distributed systems and fault tolerance #FOS: Computer and information sciences #Networking and Internet Architecture (cs.NI) #Parallel Computing and Optimization Techniques
paper · pdf · doi:10.48550/arxiv.cs/0606076
openalex publication_date 2006/06/16 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
In grid networks, distributed resources are interconnected by wide area\nnetwork to support compute and data-intensive applications, which require\nreliable and efficient transfer of gigabits (even terabits) of data. Different\nfrom best-effort traffic in Internet, bulk data transfer in grid requires\nbandwidth reservation as a fundamental service. Existing reservation schemes\nsuch as RSVP are designed for real-time traffic specified by reservation rate,\ntransfer start time but with unknown lifetime. In comparison, bulk data\ntransfer requests are defined in terms of volume and deadline, which provide\nmore information, and allow more flexibility in reservation schemes, i.e.,\ntransfer start time can be flexibly chosen, and reservation for a single\nrequest can be divided into multiple intervals with different reservation\nrates. We define a flexible reservation framework using time-rate function\nalgebra, and identify a series of practical reservation scheme families with\nincreasing generality and potential performance, namely, FixTime-FixRate,\nFixTime-FlexRate, FlexTime-FlexRate, and Multi-Interval. Simple heuristics are\nused to select representative scheme from each family for performance\ncomparison. Simulation results show that the increasing flexibility can\npotentially improve system performance, minimizing both blocking probability\nand mean flow time. We also discuss the distributed implementation of proposed\nframework.\n