CAREER: Scalable, High Performance Network Simulations Using Reverse Computation

互联网数据流量每年翻一番。如果这一速度继续下去，数据流量将在2002年左右超过语音流量。然而，网络电话和移动的支持网络的PDA产生的潜在数据没有包括在互联网流量增长率估计中，未来不可预见的"杀手级应用程序"的影响也没有包括在内，这些应用程序可能会大大增加当前对带宽的需求。因此，数据增长率在总体上可能会明显更高，至少，一些“热点”可能会经历每年四倍的流量。 不幸的是，由于技术障碍，每根光纤的带宽增长率将被限制在每年仅翻一番。因此，可能导致可用带宽的短福尔斯下降，从而将有效管理带宽的负担置于过度工作的网络管理团队上。网络管理人员需要技术和工具，使他们能够"nowcast"不仅是他们的本地网络，但周围的网络，以及为了确保，稳定，有效的带宽分配在面对动态，高带宽，下一代"杀手网络应用程序"。为了解决这个"nowcasting"问题，我们提出了一个新的并行仿真建模技术称为"反向计算"的使用。在这里，为并行执行设计的网络模型能够在模拟时间内向前和向后执行。对于简单的网络模型，反向计算已被证明可以减少并行optimiticsimulations的100倍的状态存储器的要求，并增加了6倍的整体加速比相比，经典的状态保存技术，用于支持回滚处理在乐观的模拟。我们也相信反向计算将允许大规模网络模型扩展到更大的处理器配置，以及使仿真实验的设计更有效。这个项目的总体目标是了解反向计算在应用于大规模系统建模时的基本功能和性能限制。由于其重要性和影响力，我们选择了网络模型作为我们的驱动应用，为了实现这一目标，我们提出在以下五个主要研究方向研究逆向计算：1.设计和实现用于事件列表管理和时间管理的完全可逆计算算法，以实现可扩展、高效的乐观事件处理，2.在并行仿真、逆向计算框架中有效地对大规模、多协议网络场景进行建模的过程和技术的发展;反向计算性能与保守同步技术的比较; 4.利用逆向计算设计模拟实验的新方法和新技术的创造; 5.探索反向计算、量子计算和经典并行/分布式计算之间的联系，这可能会导致对这些不同类别的计算有一个更统一的看法。

Internet data traffic is doubling each year. If this rate continues,data traffic will surpass voice traffic around the year 2002. However,not included in the Internet traffic growth rate estimates are thepotential data generated by web phones and mobile web-enabled PDAs,nor are the effects of future, unforeseen ``killer apps'' that maygreat increase the current demand for bandwidth. Thus, the data growthrate could be significantly higher in the aggregate and at the veryleast, some ``hot sites'' may experience a quadrupling of traffic eachyear. Unfortunately, due to technological barriers, bandwidth growthrates per fiber will be limited to only doubling peryear. Consequently, short falls in available bandwidth may result,thus placing the burden to effectively manage bandwidth on overworked,network management teams. Network managers will require techniques andtools that enable them to "nowcast" not only their local network, butsurrounding networks as well in order to ensure, stable, effectivebandwidth allocation in the face of dynamic, high-bandwidth, nextgeneration ``killer web apps''.To address this "nowcasting" problem, we propose the use of a newparallel simulation modeling technique called "reversecomputation". Here, network models designed for parallel execution areable to execute both forwards and backwards in simulated time. Forsimplistic network models, reverse computation has been shown toreduce the state memory requirements of parallel optimisticsimulations by a factor of 100 and increase the overall speedup by afactor of 6 when compared to classic state-saving techniques used tosupport rollback processing in optimistic simulations. We also believereverse computation will allow large-scale network models to scale tomuch larger processor configurations as well as enable a moreefficient design of simulation experiments.The overall goal of this project is to understand the fundamentalfunctional and performance limits of reverse computation when appliedto the modeling of large-scale systems. Because of its importance andimpact, we have selected network models as our driving application.To achieve this goal, we propose to investigate reverse computation inthe following five major research thrust areas:1. the design and implementation of perfectly reversible computationalgorithms for event-list management, and time management to enablescalable, efficient optimistic event processing,2. the development of processes and techniques to effectively model alarge-scale, multi-protocol network scenario in a parallel simulation,reverse computation framework,3. the comparison and contrast of reverse computation performance tostate-of-the-art conservative synchronization techniques,4. the creation of new methods and techniques for the design ofsimulation experiments that take advantage of reverse computation,and5. the exploration of the linkages between reverse computation,quantum computing and classic parallel/distributed computing thatcould lead to a more unified view of these disparate classes ofcomputation.