云计算环境中P2P计算的优化组织模型

Optimal organization model of P2P computing in the cloud computing era

P2P计算具有一些良好的特质,但是由于资源分布的任意性、互联网较大的延时、较低的有效带宽以及较高的数据传输代价,实际中目前的P2P计算效率受到了很大限制。对此问题以前的研究工作主要集中于在现行的P2P计算模型下,对一些机制进行改进。与这些先前工作不同,该研究提出了一个新的模型,它对P2P计算进行优化的组织,即将一些合适的P2P计算实例调度到适合的云计算节点上,并且以最优的方式来为其调度所需资源。更重要地,该文对优化组织过程进行了详细的数学分析和深入的理论建模,并同时对性能和代价进行了考虑。由于问题的解空间将随着问题规模的增长以指数速度扩张,因此提出了基于生物免疫思想的智能计算方法。实验验证了该算法的有效性和效率;并且与现行P2P计算模式进行对比实验的结果表明:该模型和方法为不同类型的P2P计算任务节省了运行时间和实际代价。

P2P computing provides many benefits; however, the P2P computing efficiency is limited due to the randomicity of the resource distribution, the Internet＇s inherent relatively large latency, low bandwidth, and relatively high data transfer cost. Unlike previous works to improve just some aspects of P2P computing in the current style, this paper changes the current computing style to optimize the P2P computing by migrating some peers＇ computations to computing nodes that have just been born in the emerging cloud computing era. More importantly, this paper formalizes the optimal organization process with a suite of analytical optimization models for both performance and cost. The optimization solution space expands exponentially as the problem scale increases, so this work gives a heuristic for the solution. Experiments demonstrate the effectiveness and efficiency of the heuristic and extensive comparative test results show that these models substantially reduce both the total time and the total cost.