Academic
Publications
NapSAC: design and implementation of a power-proportional web cluster

NapSAC: design and implementation of a power-proportional web cluster,10.1145/1851290.1851294,Andrew Krioukov,Prashanth Mohan,Sara Alspaugh,Laura Keys

NapSAC: design and implementation of a power-proportional web cluster   (Citations: 5)
BibTex | RIS | RefWorks Download
Energy consumption is a major and costly problem in data centers. A large fraction of this energy goes to powering idle machines that are not doing any useful work. We identify two causes of this inefficiency: low server utilization and a lack of power-proportionality. To address this problem we present a design for an power-proportional cluster consisting of a power-aware cluster manager and a set of heterogeneous machines. Our design makes use of currently available energy-efficient hardware, mechanisms for transitioning in and out of low-power sleep states, and dynamic provisioning and scheduling to continually adjust to workload and minimize power consumption. With our design we are able to reduce energy consumption while maintaining acceptable response times for a web service workload based on Wikipedia. With our dynamic provisioning algorithms we demonstrate via simulation a 63% savings in power usage in a typically provisioned datacenter where all machines are left on and awake at all times. Our results show that we are able to achieve close to 90% of the savings a theoretically optimal provisioning scheme would achieve. We have also built a prototype cluster which runs Wikipedia to demonstrate the use of our design in a real environment.
Conference: ACM SIGCOMM Conference - SIGCOMM , pp. 15-22, 2010
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.
    • ...Examples include consolidating load onto a small number of servers, e.g., using request redirection [8, 17] or...
    • ...To see why, consider two common applications: a simple cluster-based web server [8, 17] and an Hadoop cluster [20]...
    • ...Examples include designing platforms that balance CPU and I/O capacity [5, 32], routing requests to locations with the cheapest energy [29], and dynamically activating and deactivating nodes as demand rises and falls [8, 17, 37]...

    Navin Sharmaet al. Blink: managing server clusters on intermittent power

    • ...Prior studies [5] have shown that the request rate can vary significantly, with the average CPU utilization for most data center servers varying between 10% and 50% of the peak utilization...
    • ...NapSac [5] demonstrates the benefits of using a centralized policy for predicting workloads and putting machines to sleep in a heterogeneous cluster composed of Atom and Xeon nodes...
    • ...Minimizing the power spent on idle servers by keeping the minimum number of servers awake for handling spikes would improve energy efficiency [5]...

    M. Mustafa Rafiqueet al. Power management for heterogeneous clusters: An experimental study

    • ...The authors in [7] are proposing a mixture of high-end Xeon servers combined with low-end mobile processors in order to achieve a fine granularity of system capacity in relation to the workload...
    • ...Similarly to [7] the CPU cores are switched on and off according to the workload to give a better power proportionality of the system...
    • ...Related experiments in [7] result also in the capacity of 5 requests per second for the BeagleBoard which, as stated, would compare to a file size of 248 KB...

    Simon Holmbackaet al. Power proportional characteristics of an energy manager for web cluste...

    • ...Krioukov et al. [24] present a design of a power-proportional cluster, in which a cluster manager runs a knapsack algorithm in order to provision an optimal number of servers...

    Juan M. Tiradoet al. Predictive Data Grouping and Placement for Cloud-Based Elastic Server ...

    • ...Each of the server types exhibits a dierent service rate and energy consumption footprint [6]...

    Ananth Narayan Sankaranarayananet al. Global cost diversity aware dispatch algorithm for heterogeneous data ...

Sort by: