Minimizing Power Consumption with Performance Efficiency Constraint in Web Server Clusters
Minimizing Power Consumption with Performance Efficiency Constraint in Web Server Clusters
复制标题
通过 Web 服务器集群中的性能效率约束最大限度地降低功耗
DOI：
10.1109/nbis.2009.101
复制
发表时间：
2009
期刊：
影响因子：
--
通讯作者：
M. Takizawa
中科院分区：
文献类型：
--
作者：
Yan Yang;N. Xiong;A. Aikebaier;T. Enokido;M. Takizawa
能源效率已成为减少地球上资源消耗的一个非常重要的问题。我们必须考虑如何随时随地节省能源消耗，包括大量的服务器，例如Google公司的服务器。因此，电力和能源消耗最近已成为关键问题，特别是大量的服务器部署在大型集群配置中，如在数据中心和Web托管设施中。尽管我们强调尽可能地节能，但服务器的性能应该得到保证。到目前为止，已经有一些关于加强节能的讨论。然而，据我们所知，还没有一个数学模型的消耗功率和服务器的性能进行了讨论。因此，在本文中，我们试图最大限度地减少Web服务器的性能约束的能源消耗。我们知道，响应时间是Web服务器性能的焦点。因此，我们首先根据优化方法找到每个服务器的负载权衡值。通过权衡负载，可以在最小化功耗的同时满足用户的性能。然后，利用负载折衷值，提出了一种新的负载分配方法，并通过与随机负载分配方法和平均负载分配方法的比较，讨论了该方法的有效性。
Energy efficiency has become a very important issue to reduce the consumption of resources on the Earth. We have to consider how to save the energy consumption anywhere and anytime, including a large set of servers, for example, servers in the Google company. Hence, power and energy consumption has recently become key concerns, especially huge number of servers are deployed in large cluster configurations as in data centers and Web hosting facilities. Even though we emphasize power saving as much as possible, the performance of servers should be ensured. So far, there have been some discussions about enhancing power conservation. However, with the best of our knowledge, a mathematical model about consumed power and server’s performance has not been discussed. Therefore, in this paper, we attempted to minimize the energy consumption with performance constraint of web servers. We know, the response time is the focus of the performance in web servers. Therefore, we first find the trade off value of load for each server based on optimized methods. With the trade off load, the performance of users can be satisfied at the same time the power consumption is minimized. Then, using the trade off load value, we propose a novel load allocation method and then discuss the effectiveness of our load allocation method by comparing with other two methods: random load allocation method and average load allocation method.