Optimal power allocation in server farms

Abstract

Server farms today consume more than 1.5% of the total electricity in the U.S. at a cost of nearly $4.5 billion. Given the rising cost of energy, many industries are now seeking solutions for how to best make use of their available power. An important question which arises in this context is how to distribute available power among servers in a server farm so as to get maximum performance. By giving more power to a server, one can get higher server frequency (speed). Hence it is commonly believed that, for a given power budget, performance can be maximized by operating servers at their highest power levels. However, it is also conceivable that one might prefer to run servers at their lowest power levels, which allows more servers to be turned on for a given power budget. To fully understand the effect of power allocation on performance in a server farm with a fixed power budget, we introduce a queueing theoretic model, which allows us to predict the optimal power allocation in a variety of scenarios. Results are verified via extensive experiments on an IBM BladeCenter. We find that the optimal power allocation varies for different scenarios. In particular, it is not always optimal to run servers at their maximum power levels. There are scenarios where it might be optimal to run servers at their lowest power levels or at some intermediate power levels. Our analysis shows that the optimal power allocation is non-obvious and depends on many factors such as the power-to-frequency relationship in the processors, the arrival rate of jobs, the maximum server frequency, the lowest attainable server frequency and the server farm configuration. Furthermore, our theoretical model allows us to explore more general settings than we can implement, including arbitrarily large server farms and different power-to-frequency curves. Importantly, we show that the optimal power allocation can significantly improve server farm performance, by a factor of typically 1.4 and as much as a factor of 5 in some cases.

Computing Resource Allocation in Three-Tier IoT Fog Networks: A Joint Optimization Approach Combining Stackelberg Game and Matching

“…which consists of the cost caused by the queuing delay o j at the servers as well as the cost caused by the network delay h j from the sensors to the physical service provider and from the physical service provider to DSS s j . According to the queuing delay model in [5], [7], which can be easily extended to other models, the cost of queuing delay when serving DSS s j is…”

SizeCap: Efficiently handling power surges in fuel cell powered data centers

“…In order to determine how the provisioned power impacts future workload performance, the power budget planner of C-FCA-WA relies on the workload power utility function ( f utility ), which characterizes the workload performance under different server power budgets and workload intensities. 6 The power budget planner collects the future workload intensity estimates for all servers within the rack ({λ i (t + k) , k = 1, ..., P}). Based on this information, for each power capping period in the near future, the power budget planner uses the average server power budget ( 1 N P Rack (t + k)) and the average workload intensity within the rack ( 1 N ∑ N i=1 λ i (t + k)) to approximate the average workload performance f utility within the rack.…”

Power-Aware Virtual Network Function Placement and Routing Using an Abstraction Technique

“…Many studies have shown a massive waste of power consumed by idle or under-utilized devices. On one hand, PMs that are utilized by VNF instances are not power-proportional, since they consume more than 50% of their maximum power when they are in idle state [5]. Therefore, to provide powerefficiency, one solution is to increase PMs resource utilization, which leads to using lower number of active PMs.…”