包簇框架云资源分配规划

CLUSTER FRAMEWORK CLOUD RESOURCES ALLOCATION PLANNING

大量云计算数据中心建立的同时,维持数据中心的正常运转产生了巨大能耗。针对高能耗的问题,提出一种包簇框架下服务器预留分配规划算法。该算法不再以传统虚拟机为中心的资源分配方式作为模型,而是提出全新包簇框架模型下实现虚拟机到服务器的映射。利用中心极限定理来确定要预留服务器的数量,通过恒定数量的服务器从闲置状态转换为待机状态。这样不仅有效减少了服务器开关机的次数,而且能第一时间满足用户需求。资源分配规划运用0-1背包算法,将同一级别的包映射到同一级别的簇,该过程递归重复,直到所有包映射到簇。实验表明,该算法不仅降低了能耗,还提高了资源利用率,起到有效节能的作用。

While building a large number of cloud computing data centers, maintaining the normal operation of the data center has generated a lot of energy consumption. To solve the problem of high energy consumption, we proposed a server reservation allocation planning algorithm based on clustering framework. The algorithm no longer uses the traditional virtual machine-centric resource allocation method as a model. Instead, it proposes a new cluster framework model to map virtual machines to servers. The central limit theorem was used to determine the number of servers to be reserved, and a constant number of servers were converted from idle to standby. This not only effectively reduced the number of server turn-on and turn-offs, but also met the need of user in a timely manner. Resource allocation planning used 0-1 knapsack algorithm to map packets of the same level to clusters of the same level, and the process repeated recursively until all packets are mapped to the cluster. Experimental results show that the algorithm not only reduces the energy consumption, but also improves the utilization of resources, and plays an effective role in energy saving.