Analysis of the Power and Hardware Resource Consumption of Servers under Different Load Balancing Policies

Most Internet applications employ some kind of load balancing policies in a cluster setting to achieve reliable service provision as well as to deal with a resource bottleneck. However, these policies may not ensure the utilization of \textit{all} of the hardware resources in a server equally efficiently. This paper experimentally investigates the relationship between the power consumption and resource utilization of a multimedia server cluster when different load balancing policies are used to distribute a workload. Our observations are the following: (1) A bottleneck on a single hardware resource can lead to a significant amount of underutilization of the entire system. (2) A ten times increment in the network bandwidth of the entire cluster can double the throughput of individual servers. The associated increment in power consumption of the individual servers is 1.2% only. (3) For TCP-based applications, session information is more useful than other types of status information to utilize power more efficiently. (4) The use of dynamic frequency scaling does not affect the overall throughput of IO-bound applications but reduces the power consumption of the servers; but this reduction is only 12% of the overall power consumption. More power can be saved by avoiding a resource bottleneck or through service consolidation.

[1]  Amin Vahdat,et al.  Managing energy and server resources in hosting centers , 2001, SOSP.

[2]  Jeffrey S. Chase,et al.  Making Scheduling "Cool": Temperature-Aware Workload Placement in Data Centers , 2005, USENIX Annual Technical Conference, General Track.

[3]  E. N. Elnozahy,et al.  Energy-Efficient Server Clusters , 2002, PACS.

[4]  Jerome A. Rolia,et al.  Resource pool management: Reactive versus proactive or let's be friends , 2009, Comput. Networks.

[5]  S. Huang,et al.  Energy-Efficient Cluster Computing via Accurate Workload Characterization , 2009, 2009 9th IEEE/ACM International Symposium on Cluster Computing and the Grid.

[6]  Yahya Slimani,et al.  Dynamic Load Balancing Strategy for Grid Computing , 2006 .

[7]  Feng Zhao,et al.  Energy aware consolidation for cloud computing , 2008, CLUSTER 2008.

[8]  Luiz André Barroso,et al.  The Case for Energy-Proportional Computing , 2007, Computer.

[9]  Alexander Schill,et al.  Energy-aware service execution , 2011, 2011 IEEE 36th Conference on Local Computer Networks.

[10]  Suman Nath,et al.  Energy-Aware Server Provisioning and Load Dispatching for Connection-Intensive Internet Services , 2008, NSDI.

[11]  Philip S. Yu,et al.  Dynamic Load Balancing on Web-Server Systems , 1999, IEEE Internet Comput..

[12]  Keith W. Ross,et al.  A Measurement Study of a Large-Scale P2P IPTV System , 2007, IEEE Transactions on Multimedia.