Evaluating energy saving system of data centers based on AHP and fuzzy comprehensive evaluation model

Due to the high energy consumption of communication, energy saving of data centers must be enforced. But the lack of evaluation mechanisms has restrained the process on energy saving construction of data centers. In this paper, energy saving evaluation index system of data centers was constructed on the basis of clarifying the influence factors. Based on the evaluation index system, analytical hierarchy process was used to determine the weights of the evaluation indexes. Subsequently, a three-grade fuzzy comprehensive evaluation model was constructed to evaluate the energy saving system of data centers.


1.
Introduction Energy consumption of data centers accounts the main part of that of the communication network. Most of the old data centers have low energy efficiency because of poor design and technical level. Therefore, energy saving work of data centers is imminent. Scientific evaluation of energy saving of data centers is an important basis for carrying out communication energy conservation. Therefore, it is of great significance to establish an energy saving evaluation system of data centers.
Energy saving standards of data centers in developed countries have begun to take shape [1][2][3]. But different national conditions result in a big difference in evaluation indexes [4,5]. And the suitable model to evaluate the energy saving system of data centers has never been reported till now.
This work tries to establish an energy saving evaluation index system of data centers of China. In addition, analytic hierarchy process and the fuzzy comprehensive evaluation method are used to evaluate this energy saving system of data centers. Finally, the validity of the model is verified by case study.

Constructing the evaluation index system
In this work, two first grade indexes, six second grade indexes and twenty-one third grade indexes are selected to construct the evaluation system. Green energy saving assessment D 244 Improvement and promotion D 245

Constructing judgment matrix ( A )
The analytic hierarchy process method is applied to determine the weights of the various factors in The term ij a represents the relative importance of i a compared to j a , the value of ij a is determined by a 9-point scale [6].

Calculating the weights of evaluation indexes
In this work, six experts were invited to evaluate the impact of each index on the operation ability. Then the specific score of every indicator was gained on the basis of a percentage grading system. The results are shown in Table 2.

Establishment of comments set
The comments set indicates different grades from low to high to reflect the performance of the evaluated object of various indexes. We can determine comments set as: ={excellent, good, moderate, poor, bad}. The comments set and the responding scores for energy saving system of data centers are shown in Table 3. Table 3 The comments set and the responding scores for energy saving system of data centers.

Determination of weight set
According to the weights of each evaluation index in Table 2, the weight set can be determined. For example,  

Establishment of the membership function
The methods to determine membership function are varied [7]. In this paper, the membership function is selected as follows:

Determination of the fuzzy relationship matrix based on the membership function
(1) Comments set of single factor Comments set of single factor consists of the membership of this factor belonging to the The fuzzy comprehensive evaluation matrix ij R of index ij C can be defined as follows: where m is the number of the index of ij C . ijm r represents the membership of ijk D belonging to

Obtaining the comprehensive evaluation vector
Comprehensive evaluation vector can be obtained by synthesizing the judgment matrix with the weight matrix using a suitable operator.
where Q is the comprehensive evaluation vector, P is the weight matrix, R is the judgment matrix. '  ' is a weighted averaging operator.
Finally, on the basis of maximum membership degree principle, the evaluation grade of the evaluated object can be determined.

Evaluation object
This data center built in 2012 has an area of 205m 2 . It has 18 switching equipments, 66 servers and 6 dedicated air conditioners, a dedicated electricity meter, efficient UPS system and video surveillance system. The building layout and equipment configuration reflect energy saving obviously. The data center also applied air flow organization optimization, air-conditioning inverter technology, high voltage DC power supply and other energy saving technology.
According to the actual situation of this data center, the indexes scores given by experts are shown in Table 4.