- Research
- Open Access

# A big data placement method using NSGA-III in meteorological cloud platform

- Feng Ruan
^{1}, - Renhao Gu
^{2}Email author, - Tao Huang
^{4}and - Shengjun Xue
^{2, 3, 4}Email author

**2019**:143

https://doi.org/10.1186/s13638-019-1456-7

© The Author(s) 2019

**Received:**15 March 2019**Accepted:**25 April 2019**Published:**3 June 2019

## Abstract

Meteorological cloud platforms (MCP) are gradually replacing the traditional meteorological information systems to provide information analysis services such as weather forecasting, disaster warning, and scientific research. However, the explosive growth of meteorological data resources has brought new challenges to the placement and management of big data in MCP. On the one hand, managers of MCP need to save energy to achieve cost savings. On the other hand, users need shorter data access time to improve user’s experience. Hence, a big data placement method in MCP is proposed in this paper to deal with challenges above. First, the resource utilization, the data access time, and the energy consumption in MCP with the fat-tree topology are analyzed. Then, a corresponding data placement method, using the improved non-dominated sorting genetic algorithm III (NSGA-III), is designed to optimize the resource usage, energy saving, and efficient data access. Finally, extensive experimental evaluations validate the efficiency and effectiveness of our proposed method.

## Keywords

- Meteorological cloud platform
- Big data
- Data placement
- NSGA-III

## 1 Introduction

The meteorological information system is a system which is used to store and analyze a number of the meteorological professional work via some certain relationships [1]. Meteorological information system requires that meteorological information be stored in the form of computer software system [2]. The service it provides covers from data collection, retrieval to processing, analysis, and prediction, including forecasters’ understanding of computer weather products, decisions, and visual submission of forecast conclusions, to form a complete system workflow [3]. It can effectively analyze meteorological conditions and carry out meteorological forecasting and disaster warning, which are closely related to people’s production activities and daily life [4].

However, with the daily growth rate of 12 TB, the traditional meteorological information system has been difficult to deal with the explosive growth of meteorological data resources [5, 6]. The construction of meteorological cloud platform (MCP) supported by big data is imminent. MCP has irreplaceable advantages in business, service, scientific research, and government affairs [7]. Its data resources have the characteristics of complete data, high quality, and fast usage [8]. Relying on the public cloud platform, it can provide personalized services such as big data analysis [9].

When it comes to MCP, cloud computing technology can be well applied to such scenarios. Cloud computing is a style of computing in which dynamically scalable and often virtualized resources are provided as a service over the Internet [10]. Cloud computing has the characteristics of super-large-scale virtualization, high reliability, and so on [11, 12]. It can process these massive meteorological resource data in a timely and batch manner [13]. By accessing meteorological information through parallel network, users’ requests for meteorological analysis tasks can be processed efficiently [14, 15].

Nevertheless, when facing with the mountains of the meteorological information, it contributes to the problem that the data management of MCP becomes difficult [16]. On the one hand, the energy consumption of MCP is greatly increased in order to execute the meteorological task requests, which results in a great cost for the managers of the MCP [17]. On the other hand, how to solve the task requests accurately and timely is another problem of the MCP, which is closely related to the Quality of Service [18]. Hence, the MCP needs a data placement method that takes into account both energy consumption and data access time [19].

Although the energy-efficient data placement methods of the meteorological information system have been studied in many researches, few of them took the resource utilization of MCP, the data access time of the meteorological tasks, and the energy consumption of MCP into consideration. With the observation above, it is still a challenge to realize the big data placement in MCP. In view of this challenge, a big *d*ata *p*lacement *m*ethod in *M*CP, named DPMM, is proposed in this paper.

The main contributions of this paper are listed as follows:

(1) Analyze the resource utilization, the data access time, and the energy consumption model in MCP and formulate the multi-objective problem of the data placement.

(2) Employ improved non-dominated sorting genetic algorithm III (NSGA-III) to improve the resource utilization of MCP and reduce the data access time of the tasks and the energy consumption in MCP.

(3) Conduct adequate experimental evaluation and comparison analysis to validate the efficiency and effectiveness of our proposed method DPMM.

The rest of this paper is organized as follows. In Section 2, the model of MCP which employed the fat-tree topology is built. In Section 3, formalized concepts, systematic models with the resource utilization, data access time, and the energy consumption which are taken into account, and the multi-objective problem are presented. The proposed data placement in MCP is detailed in Section 4. Section 5 illustrates the comparison analysis and performance evaluation. The related work is summarized in Section 6, and Section 7 concludes this paper with the future work presented.

## 2 Meteorological cloud platform over big data

In meteorological cloud platforms, meteorological data are distributed on thousands of physical machines. When a computational task arrives at a physical machine, it may require a variety of meteorological data at different locations for computation. Therefore, this paper uses fat-tree topology structure to construct the meteorological database of MCP [20].

The advantages of the fat-tree structure are obvious. First of all, its bandwidth division increases with the expansion of network scale which can provide high throughput transmission service for MCP. In addition, it can provide communication between different pods. Moreover, there are many parallel paths between source and destination nodes, the fault-tolerant performance of the network is guaranteed, and there will be no single point of failure in general. Besides, the fat-tree structure achieves timely load handling through multiple links at the core layer, avoiding network hot spots, and avoids overload by reasonable shunting within the pod.

The structure of the fat-tree is mainly composed of three layers of switches which are the core switch, the aggregation switch, and the edge switch, and the edge switch is connected with the physical machine. From our former research [21], it can be seen that in the fat-tree topology, every two aggregation switches and two edge switches form a pod. If we figure out the number of pods, the whole network topology is clear. Assuming that the number of the pods in MCP is *p*, then the number of physical machines that each pod can connect to is (*p*/2)^{2}, and the number of edge switches and aggregate switches in each pod is *p*/2; there are (*p*/2)^{2} core switches in the whole topology. The number of ports per switch in the network is *p*, thus the topology can be connected to *p*^{3}/4 physical machines in total.

## 3 Resource model of MCP

In this section, we construct a resource model of MCP in big data environment based on the fat-tree network topology structure and establish the multi-objective problem of big data distribution in MCP [22].

### 3.1 System resource model

In this paper, we focus on a big data-oriented MCP that provides flexible resources for storing meteorological data. Suppose that there are *N* meteorological datasets in the current MCP that need to be placed on the physical machines, denoted as *D*={*d*_{1},*d*_{2},…,*d*_{N}}, and *M* tasks have arrived at the MCP to be performed, denoted as *T*={*t*_{1},*t*_{2},…,*t*_{M}}. A task may require multiple meteorological datasets to perform, and a meteorological dataset may also be accessed by multiple tasks. Since the number of pods is *p* in the MCP, thus there are *p*^{3}/4 physical machines provided to place the datasets and the tasks. Assuming *W* is the total number of physical machines, *W*=*p*^{3}/4, then the physical machines in the MCP can be represented by *PM*={*pm*_{1},*pm*_{2},…,*pm*_{W}}. Hence, set *X*={*x*_{1},*x*_{2},…,*x*_{N}} be the collection of the placement strategy for each meteorological dataset, and each placement strategy *x*_{n}∈*PM*(1≤*n*≤*N*). Similarly, set *Y*={*y*_{1},*y*_{2},…,*y*_{M}} be the collection of the placement location of each task set, and each placement location *y*_{m}∈*PM*(1≤*m*≤*M*). Therefore, this paper will solve the placement strategy *X* of meteorological dataset according to the known task placement strategy *Y*. Meteorological data refers to information elements or numerical analysis results collected or processed by all possible means of observation, detection, and telemetry, which come from the Earth’s atmosphere and its adjacent layers and are related to the law of atmospheric state change.

*x*

_{n}=

*y*

_{m}. Using binary flag

*ξ*

_{n}to represent if

*d*

_{n}is the industrial social data, thus

### 3.2 Resource utilization model

In MCP, task sets and datasets are managed by virtual machines placed on physical machines. The resources required by datasets and task sets and the capacity of physical machines can be expressed by the number of virtual machine instances. Suppose *c*_{w} be the capacity of the *w*th physical machine *pm*_{w}.

*X*of meteorological datasets, the resource utilization of each physical machine can be obtained by detecting the resource utilization of virtual machine instances. Suppose

*u*

_{w}(

*X*) be the resource utilization of the

*w*th physical machine under the placement strategy

*X*, which can be calculated by

*θ*

_{n}represents the resource demand of meteorological dataset

*d*

_{n}, and

*I*

_{n,w}(

*X*) is the binary flag to judge whether

*d*

_{n}is placed on

*pm*

_{w}, thus

*λ*(

*X*) be the number of employed physical machines, which can be calculated by

*F*

_{w}(

*X*) is the binary flag to judge whether

*pm*

_{w}is employed, thus

### 3.3 Access time model

In the MCP, the task set needs to be implemented by the meteorological dataset deployed on the physical machine. Hence, when considering the big data distribution strategy of meteorological cloud platform, it is necessary to consider the data access time of task set accessing meteorological dataset.

*l*

_{m,n}is the binary flag to judge whether

*t*

_{m}needs to access

*d*

_{n}, thus

*g*

_{m,n}is the access frequency of

*t*

_{m}access to

*d*

_{n}in a task execution cycle

*R*, then the total access frequency

*K*in a period of execution cycle can be calculated by the following formula

*ES*(

*pm*) be the edge switch connected to the physical machines and

*Pod*(

*pm*) represents the pod of this edge switch. Then, the number of exchanges in data transmission can be divided into four cases, respectively: (1)

*t*

_{m}and

*d*

_{n}are placed on the same physical machine, which means

*x*

_{n}=

*y*

_{m}. (2)

*t*

_{m}and

*d*

_{n}are placed on the different physical machines but connected to the same edge switch, which means

*x*

_{n}≠

*y*

_{m},

*ES*(

*x*

_{n})=

*ES*(

*y*

_{m}). (3) The physical machines of

*t*

_{m}and

*d*

_{n}belongs to different edge switches but the same pod, which means

*ES*(

*x*

_{n})≠

*ES*(

*y*

_{m}),

*Pod*(

*x*

_{n})=

*Pod*(

*y*

_{m}). (4) The physical machines of

*t*

_{m}and

*d*

_{n}belong to different pods, which means

*Pod*(

*x*

_{n})≠

*Pod*(

*y*

_{m}). Based on the analysis above, the access time

*t*

_{m,n}(

*X*) of

*t*

_{m}to access

*d*

_{n}once can be calculated by

*ρ*

_{n}represents the data volume of

*d*

_{n}and

*b*

_{P}

*E*represents the bandwidth between physical machines and edge switches; similarly,

*b*

_{E}

*A*represents the bandwidth between edge switches and aggregation switches and

*b*

_{A}

*C*represents the bandwidth between aggregation switches and the core switches. Then, the average data access time

*T*(

*X*) of the MCP can be calculated using the following formula

### 3.4 Energy consumption model

where *α*_{w} is the basic energy rate of *pm*_{w}.

*η*is the active energy rate of the virtual machines. And the virtual machine instances that are not occupied by meteorological datasets will consume less energy, which we call it the idle virtual machine consumption, denoted as

*IE*, can be calculated by

where *τ* is the idle energy rate of the virtual machines.

*p*

^{2}, in addition, the number of switches passed by

*t*

_{m}to access

*d*

_{n}, denoted as

*δ*

_{m,n}(

*X*), can be calculated by

*SE*can be calculated by

where *β* is the basic energy rate of the switches and *γ* is the working energy rate of the switches.

### 3.5 Problem formulation

The focus of our research is to solve the problem of big data placement in the environment of MCP. On the basis of known task set placement strategy, we optimize the placement strategy of meteorological dataset in order to improve resource utilization, reduce data access time and energy consumption.

## 4 A big data placement method in MCP

In this section, a data placement method in MCP is proposed. Compared with the other data placement, NSGA-III is widely applied in could platform because of its advantages in finding the optimal solution in the feasible solution accurately and timely, which is used to MCP in this paper [23, 24]. First, the placement strategies for meteorological data sets are encoded and fitness functions are given for the optimization problem. Second, the fast non-dominated sorting approach and the crowded-comparison operation are used in selection. Then, the crossover and mutation operation of traditional genetic algorithm (GA) are adopted. Finally, the overview of our method is described in detail.

### 4.1 Data placement based on NSGA-III

A data placement strategy in MCP is proposed in this subsection. The data placement problem is defined as a multi-objective optimization problem. NSGA-III is an accurate and robust method for addressing the optimization problem with multiple objectives. Here, NSGA-III is adopted to solve the multi-objective optimization problem presented in [18].

Firstly, we encode for the physical machines and then give the fitness function and constraints for the multi-objective optimization problem. In addition, we generate new chromosomes using the crossover and mutation operations. At last, we pick out the chromosomes of the next generation, using usual domination principle and reference-point-based selection. The overview of the multi-objective data placement is elaborated.

#### 4.1.1 Encoding

In this section, we encode for the physical machines and meteorological datasets are placed to the physical machines, as is presented in Section 3. In the genetic algorithm (GA), a gene represents the data placement strategy of a dataset and the genes compromise a chromosome, representing the placement strategies of all the meteorological datasets. The physical machines are encoded as 1,2,…,*W*.

*N*meteorological datasets. In this example, the chromosome is encoded in an array of

*W*integers (1,2,…,

*W*).

#### 4.1.2 Fitness functions and constraints

The fitness functions are adopted to evaluate the solutions generated by the chromosomes. In GA, each chromosome is an individual and represents a solution of the optimization problem. The fitness functions in this paper includes three categories: the average resource utilization, the data access time, and the energy consumption, presented in [6], [10], and [17] respectively. The resource utilization should be maximized and the data access time and the energy consumption must be minimized, and we aim to achieve a balance among those three objectives.

#### 4.1.3 Initialization

In this subsection, the parameters of GA are determined, including the population size *S*, the maximum iteration *I*, the crossover possibility *p*_{c}, and the mutation possibility *p*_{m}.

Each chromosome represents the placement strategies of all the meteorological datasets, which is denoted as *X*_{i} = (*x*_{1},*x*_{2},…,*x*_{N})(*i*=1,2,…,*S*). Each gene in the chromosome stands for the placement strategy of a dataset.

#### 4.1.4 Crossover and mutation

*W*genes respectively. In this example, a crossover point is determined in advance and then the genes of two chromosomes are swapped around the point. Thus, two new chromosomes are created.

*N*genes is shown in Fig. 4. Each gene in the chromosome is changed with the equal possibility

*p*

_{m}.

### 4.2 Selection for the next generation

We select the chromosomes for the next generation in the selection phase with higher fitness values. We encode for the physical machines and each chromosome represents all the data placement strategies of the meteorological datasets. As is discussed above, the number of fitness functions for each chromosome is 3, i.e., the resource utilization, the data access time, and the energy consumption. After crossover and mutation operations are conducted, *S* chromosomes are generated and the size of population becomes 2*S*. We aim at select *S* chromosomes for the next generation.

Firstly, the three objectives are evaluated according to [6], [10], and [17]. Then, usual domination principle is conducted to sort the 2*S* solutions using the three fitness values to generate some non-dominated fronts. Suppose that there are *n* non-dominated fronts, and the datasets in the first front have the most optimal fitness values.

We select one chromosome at random from the first front to the *n*th fronts every time until *S* chromosomes have been picked out. Assume the last chosen solution is in the *f*th front and the number of the selected chromosomes in the *f*th front is *s*. If the number of solutions in the *f*th non-dominated front is *s*, that is all the individuals in the *f*th front are selected, then the selection phase finishes.

However, if part of the chromosomes in the *f*th front are chosen, then further selection is required to determine the chromosomes in the *f*th front going into the next chromosomes.

*U*

^{∗}(

*X*),

*T*

^{∗}(

*X*), and

*E*

^{∗}(

*X*) respectively. We search

*U*

^{∗}(

*X*),

*T*

^{∗}(

*X*), and

*E*

^{∗}(

*X*) in the population firstly. Then the fitness functions are updated as

*U*

_{m},

*T*

_{m}, and

*E*

_{m}represent the extreme values of the average resource utilization, the data access time, and the energy consumption respectively which are calculated by

where *W*_{U},*W*_{T}, and *W*_{E} are the weight vectors to the three functions.

*ρ*

_{U},

*ρ*

_{T}, and

*ρ*

_{E}respectively. Then, the fitness functions are normalized by

After normalization, the fitness values of the resource utilization, the data access time, and the energy consumption are between zero and one. The normalized solutions are scattered in the hyperplane composed by the three fitness axes. Each solution of the individual in the population is corresponding to a triple.

*g*subsections, and then, the number of the reference points, denoted as

*ζ*, is calculated by

*ζ* is approximately equal to *S* to guarantee each individual associate with a reference point. Then, we associate the normalized solutions with the reference points.

We sort the solutions in the *f*th non-dominated front, using the number of the associated reference points. Assume the solutions in the *f*th front are sorted into *d* levels and the reference points in the first level represents the maximum reference points. Each time selects one solution randomly from the first level until the *n* individuals have been selected.

### 4.3 Solution evaluation using SAW and MCDM

The proposed strategy aims at achieving a trade-off in optimizing the average resource utilization, the data access time of the tasks, and the energy consumption of MCP. In each population, there are *N* chromosomes and each chromosome *x*_{n} represents a hybrid data placement of *N* meteorological datasets. In addition, dynamic schedule of datasets is considered and, to select relatively better schedule solution of each placement strategy, simple additive weighting (SAW) and multiple criteria decision making (MCDM) are employed.

*n*th placement strategy as

*U*

^{max}and

*U*

^{min}represent the maximum and minimum fitness for resource utilization in the

*n*th placement strategy. Similarly, the time consumption of

*n*th placement strategy is normalized as

*T*

^{max}and

*T*

^{min}represent the maximum and minimum fitness for data access time in the

*n*th placement strategy. And the energy consumption of

*n*th placement strategy is normalized as

where *E*^{max} and *E*^{min} represent the maximum and minimum fitness for energy consumption in the *n*th placement strategy. In addition, to calculate the utility value of each solution, the weight of each objective function requires determination.

*n*th placement strategy is calculated by

*UV*(

*x*

_{n}) represents the utility value of the

*n*th placement strategy and

*κ*

_{U},

*κ*

_{T}, and

*κ*

_{E}represent the weight of the resource utilization, the data access time, and the energy consumption respectively. Therefore, for each chromosome in the population, we have calculated the utility value of the same schedule. The formalized optimization problem given in (18) can be represent by

The solution selection strategy is elaborated in Algorithm 1. We select the maximum and minimum value of the three objections (Lines 1-6). Then the utility values of the three objections and the solutions are evaluated (Lines 7-12). The maximum utility value of solution is selected (Line 13). Finally, the solution with maximum utility value is output.

### 4.4 Method overview

In this paper, we aim to optimize the resource utilization, reduce the data access time, and decrease the energy consumption, which is determined as a multi-objective optimization problem. The NSGA-III is adopted to achieve the global optimal data placement strategy. First, we encode for the physical machines and each gene represents the data placement strategy for a meteorological dataset. Then, the fitness functions and constraints of the multi-objective optimization problem are elaborated. Crossover and mutation operations are conducted to generate new chromosomes and usual domination principle and reference-point-based selection in NSGA-III are adopted to select the individuals going into the next generation.

The overview of the proposed method is elaborated in Algorithm 2. We input the tasks, the physical machines, and the placement strategy of the tasks. The algorithm starts with the first iteration, and for each iteration, we conduct the crossover and mutation operation to generate new solutions (Line 3). Then, we evaluate the fitness functions of each chromosomes and obtain the placement routing (Lines 5). We conduct the selection, including the non-dominant sorting, the primary selection, and the further selection (Lines 7-13). When the algorithm reaches the maximum iteration, we evaluate the solutions by SAW and MCDM (Line 17). Finally, the optimal data placement methods are output.

## 5 Experimental evaluation

In this section, a set of complex simulations and experiments are performed to evaluate the performance of the proposed DPMM. We first introduced the simulation setup with the simulation parameter settings and the statement of the comparative data placement. Then, the influence of different task scales on the performance of the resource utilization, the data access time, and the energy consumption of the compared method and our proposed DPMM is evaluated.

### 5.1 Simulation setup

Parameter settings

Parameter description | Value |
---|---|

The number of tasks | 40 |

The time of running period | 24 h |

The number of physical machines | 2000 |

The baseline energy consumption rate of | 342 W |

The active energy consumption rate for each virtual machine | 37 W |

The bandwidth between severs and edge switches | 50 M |

The bandwidth between switches in the fat-tree topology | 100 M |

The idle energy consumption rate for each switch | 100 W |

The active energy consumption rate for each switch | 150 W |

To conduct the comparison analysis, we employ some other basic data placement method besides our DPMM. The comparative methods are briefly expounded as follows:

• Benchmark: In this method, the meteorological datasets are placed in the order of the physical machines. When the former physical machine is full, the datasets left are placed on the next physical machine. This process is repeated until all datasets have been placed.

• First fit decreasing in MCP (FFD-M): The meteorological datasets are sorted in descending order according to the dataset requests first. Then, the sorted datasets are placed on the physical machines. If the left resources of current physical machine are insufficient for the resource requirement of a dataset, the dataset is placed on the physical machine with enough resource. This process is repeated until all datasets have been placed.

• Best fit decreasing in MCP (BFD-M): The meteorological datasets and the physical machines are both sorted in descending order according to the dataset request and the space of physical machines first. Then, the sorted datasets are placed on the sorted physical machines. If the left resources of the physical machine are insufficient for the resource requirement of the dataset, the dataset is placed on the physical machine with enough resource next to this physical machine in optimal principle. This process is repeated until all datasets have been placed.

The data placement methods are implemented on a personal computer with Intel Core i7-4720HQ 3.60 GHz processors and 4 GB RAM.

### 5.2 Performance evaluation of DPMM

The proposed DPMM is intended to achieve a trade-off with optimizing the resource utilization while reducing the data access time and the energy consumption. We conducted 50 experiments in the case of convergence for each dataset scale, and multiple sets of results are obtained.

### 5.3 Comparison analysis

In this subsection, the comparisons of Benchmark, FFD-M, BFD-M, and DPMM with the same experimental context are analyzed in detail. The resource utilization, the data access time, and the energy consumption are the main metrics for evaluating the performance of the data placement method. The corresponding results are shown in Figs. 6, 7, and 8.

(1) Comparison of resource utilization: After placing all the meteorological datasets on the physical machines via the data placement methods, the occupation of the VM instances is achieved. Figure 6 shows the comparison of the resource utilization in MCP of Benchmark, FFD-M, BFD-M, and DPMM at different dataset scales. The resource utilization is calculated according to the number of employed physical machines and the employed VM instances in each physical machine. Fewer employed physical machines with more occupied VM instances contribute to a higher resource utilization. It is intuitive from Fig. 6 that our proposed method DPMM achieves higher and stable resource utilization. That is, DPMM reduces the number of unemployed VM instances and wastes less resources than other data placement method.

(2) Comparison of data access time: Fig. 7 shows the comparison of the data access time of the meteorological tasks of Benchmark, FFD-M, BFD-M, and DPMM at different dataset scales. In this paper, we try to realize a balance between the resource utilization, the data access time and the energy consumption; however, in order to improve the resource utilization and reduce the energy consumption, the meteorological datasets may place on the physical machine which may be far from the tasks. It is illustrated from Fig. 7 that the data access time of our proposed method DPMM is a little more than the other data placement method, which means our proposed method may sacrifice some data access time to optimize the resource utilization and the energy consumption.

## 6 Related work

In recent years, with the rapid growth of meteorological data resources, people are actively seeking to replace the traditional meteorological information system platform, and the MCP has attracted extensive attention because of its high-efficiency computing ability for large data [26, 27]. Pokric et al. [6] presents the environmental monitoring solution ekoNET, developed for a real-time monitoring of air pollution and other atmospheric condition parameters such as temperature, air pressure, and humidity. In [28], Sawant et al. addresses above issues through the adaptation of a framework based on Open Geospatial Consortium standards for Sensor Web Enablement. Padarian et al. in [29] explores the feasibility of using this platform for digital soil mapping by presenting two soil mapping examples over the contiguous USA.

However, with the increasing amount of data, the problem of data placement in meteorological cloud data center is becoming more and more serious [30, 31]. Traditional data placement method cannot meet the basic requirements of query and analysis of massive data. A new data placement method for big data needs to be applied to MCP immediately [32]. The data placement method has been widely researched in many studies [33]. In [34], Yu and Pan proposed an associated data placement scheme, which improves the co-location of associated data and the localized data serving while ensuring the balance between nodes. In [35], Paiva et al. addressed the problem of self-tuning the data placement in replicated key-value stores to automatically optimize replica placement in a way that leverages locality patterns in data accesses, such that internode communication is minimized.

In the study of data placement methods, a large number of methods such as operations research are applied to optimize them [36, 37]. And a large number of studies such as [38–41] and [42] are carried out. Compared with these methods, genetic algorithms have more rapid and accurate characteristics, and better adapt to the requirements of data placement in big data environment. Many studies focus on genetic algorithms [43].

In [44], Jean-Luc et al. compared genetic algorithm to a pure random optimization approach and their optimization efficiencies are analyzed. In [45], Kadri and Boctor addressed the resource-constrained project scheduling problem with transfer times and the experiment, conducted on a large number of instances, shows that the proposed algorithm performs better than several solution methods previously published. In [46], a set-based Pareto dominance relation was then defined to modify the fast non-dominated sorting approach in NSGA-II.

However, with the observation above, few studies have been conducted about the data placement in MCP facing with the big data. In this paper, we employed one of the genetic algorithm NSGA-III to optimize the resource utilization, the data access time, and the energy consumption in MCP.

## 7 Conclusion and future work

The existing works of meteorological data placement in MCP have not combined the energy consumption and the data access time together. In this paper, a big data placement method in MCP is proposed. We construct as systematic model of the resource utilization, the data access time, and the energy consumption in MCP. The proposed method is designed to improve the average resource utilization in MCP and access performance while reducing the energy consumption using the NSGA-III. Through adequate experimental evaluation and comparison analysis, the efficiency and effectiveness of our proposed method are validated.

For future work, we will adjust and extend our method to implement the meteorological data placement in exact physical environment. Accordingly, the data access time should also be reduced than the result of this paper in future.

## Declarations

### Acknowledgements

This work is supported by The Startup Foundation for Introducing Talent of NUIST, the open project from State Key Laboratory for Novel Software Technology, Nanjing University under grant no. KFKT2017B04, the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD) fund, Jiangsu Collaborative Innovation Center on Atmospheric Environment and Equipment Technology (CICAEET), and the project “Six Talent Peaks Project in Jiangsu Province” under grant no. XYDXXJS-040.

### Funding

This research is supported by the National Science Foundation of China under grant no. 61702277, no. 61872219, and no. 61772283 with the help of data collection and analysis.

### Availability of the data and materials

The dataset employed to support the conclusion of this article is available at http://data.cma.cn/.

### Authors’ contributions

FR, RG, and TH conceived and designed the study. RG and TH performed the simulations. FR and RG wrote the paper. All authors reviewed and edited the manuscript. All authors read and approved the final manuscript.

### Authors’ information

Not applicable.

### Competing interests

The authors declare that they have no competing interests.

### Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Open Access** This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

## Authors’ Affiliations

## References

- A. Stein, R. R. Draxler, G. D. Rolph, B. J. Stunder, M. Cohen, F. Ngan, NOAA?s HYSPLIT atmospheric transport and dispersion modeling system. Bull. Am. Meteorol. Soc.
**96**(12), 2059 (2015).View ArticleGoogle Scholar - A. T. Eseye, J. Zhang, D. Zheng, Short-term photovoltaic solar power forecasting using a hybrid Wavelet-PSO-SVM model based on SCADA and Meteorological information. Renew. Energy.
**118:**, 357 (2018).View ArticleGoogle Scholar - S. Fan, J. R. Liao, R. Yokoyama, L. Chen, W. J. Lee, Forecasting the wind generation using a two-stage network based on meteorological information. IEEE Trans. Energy Convers.
**24**(2), 474 (2009).View ArticleGoogle Scholar - V. S. Sherstnyov, A. I. Sherstnyova, I. A. Botygin, D. A. Kustov, in
*Key Engineering Materials, vol. 685*. Distributed information system for processing and storage of meteorological data (Trans Tech PublBoston, 2016), pp. 867–871.Google Scholar - J Brenton, Status of the Meteorological Data Format Working Group (2016). https://ntrs.nasa.gov/search.jsp?R=20160011066. Accessed 13 May.
- B. Pokric, S. Kreo, D. Drajic, M. Pokric, I. Jokic, M. J. Stojanovic, in
*2014 Eighth International Conference on Innovative Mobile and Internet Services in Ubiquitous Computing*. ekoNET-environmental monitoring using low-cost sensors for detecting gases, particulate matter, and meteorological parameters (IEEE Computer SocietyWashington, DC, 2014), pp. 421–426.Google Scholar - J. Elston, B. Argrow, M. Stachura, D. Weibel, D. Lawrence, D. Pope, Overview of small fixed-wing unmanned aircraft for meteorological sampling. J. Atmos. Ocean. Technol.
**32**(1), 97 (2015).View ArticleGoogle Scholar - X. Xu, Y. Li, T. Huang, Y. Xue, K. Peng, L. Qi, W. Dou, An energy-aware computation offloading method for smart edge computing in wireless metropolitan area networks. J. Netw. Comput. Appl.
**133:**, 75–85 (2019).View ArticleGoogle Scholar - J. Figa-Saldaña, J. J. Wilson, E. Attema, R. Gelsthorpe, M. Drinkwater, A. Stoffelen, The advanced scatterometer (ASCAT) on the meteorological operational (MetOp) platform: A follow on for European wind scatterometers. Can. J. Remote. Sens.
**28**(3), 404 (2002).View ArticleGoogle Scholar - https://www.nice-software.com/solutions/cloud-computing. Accessed 3 Feb 2019.
- X. Wang, L. T. Yang, L. Kuang, X. Liu, Q. Zhang, M. J. Deen, A Tensor-Based Big-Data-Driven Routing Recommendation Approach for Heterogeneous Networks. IEEE Netw.
**33**(1), 64 (2018).View ArticleGoogle Scholar - S. Wang, A. Zhou, M. Yang, L. Sun, C. H. Hsu, Service composition in cyber-physical-social systems. IEEE Trans. Emerg. Top. Comput. (2017). https://doi.org/10.1109/TETC.2017.2675479.
- X. Xu, Q. Liu, Y. Luo, G. Xing, K. Peng, Z. Xu, S. Meng, L. Qi, A Computation Offloading Method over Big Data for IoT-Enabled Cloud-Edge Computing, Future Generation Computer Systems. Futur. Gener. Comput. Syst.
**95:**, 522–533 (2018).View ArticleGoogle Scholar - Y. Xu, L. Qi, W. Dou, J. Yu, Privacy-preserving and scalable service recommendation based on simhash in a distributed cloud environment. Complexity.
**2017:**, 1–9 (2017).Google Scholar - X. Wang, L. T. Yang, X. Xie, J. Jin, M. J. Deen, A cloud-edge computing framework for cyber-physical-social services. IEEE Commun. Mag.
**55**(11), 80 (2017).View ArticleGoogle Scholar - X. Xu, Y. Xue, L. Qi, Y. Yuan, X. Zhang, T. Umer, S. Wan, An edge computing-enabled computation offloading method with privacy preservation for internet of connected vehicles. Futur. Gener. Comput. Syst.
**96:**, 89–100 (2019).View ArticleGoogle Scholar - X. Xu, X. Zhang, M. Khan, W. Dou, S. Xue, S. Yu, A Balanced Virtual Machine Scheduling Method for Energy-Performance Trade-offs in Cyber-Physical Cloud Systems. Futur. Gener. Comput. Syst.
**18:**, 1–11 (2017).Google Scholar - S. Wang, A. Zhou, R. Bao, W. Chou, S. Yau, Towards Green Service Composition Approach in the Cloud. IEEE Trans. Serv. Comput. (2018). https://doi.org/10.1109/TSC.2018.2868356.
- L. Ren, X. Cheng, X. Wang, J. Cui, L. Zhang, Multi-scale Dense Gate Recurrent Unit Networks for bearing remaining useful life prediction. Futur. Gener. Comput. Syst.
**94:**, 601 (2019).View ArticleGoogle Scholar - N. Jain, A. Bhatele, L. H. Howell, D. Böhme, I. Karlin, E. A. León, M. Mubarak, N. Wolfe, T. Gamblin, M. L. Leininger, in
*Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis*. Predicting the performance impact of different fat-tree configurations (IEEE Computer SocietyWashington, DC, 2017), p. 50.Google Scholar - X. Xu, S. Fu, L. Qi, X. Zhang, Q. Liu, Q. He, S. Li, An IoT-oriented data placement method with privacy preservation in cloud environment. J. Netw. Comput. Appl.
**124:**, 148 (2018).View ArticleGoogle Scholar - X. Xu, X. Zhao, F. Ruan, J. Zhang, W. Tian, W. Dou, A. Liu, Data Placement for Privacy-Aware Applications over Big Data in Hybrid Clouds. Secur. Commun. Networks.
**2017:**, 1–15 (2017).Google Scholar - G. Wang, X. Huang, J. Zhang, Levitin–Polyak well-posedness in generalized equilibrium problems with functional constraints. Pac. J. Optim.
**6:**, 441 (2010).MathSciNetMATHGoogle Scholar - W. Gang, H. Xuexiang, Z. Jie, C. Guangya, Levitin-poyak well-posedness of generalized vector equilibrium problems with functional constraints. Acta Math. Sci.
**30**(5), 1400 (2010).MathSciNetView ArticleGoogle Scholar - http://data.cma.cn/. Accessed 26 Jan 2019.
- K. Bessho, K. Date, M. Hayashi, A. Ikeda, T. Imai, H. Inoue, Y. Kumagai, T. Miyakawa, H. Murata, T. Ohno, et al., An introduction to Himawari-8/9—Japan’s new-generation geostationary meteorological satellites. J. Meteorol. Soc. Jpn. Ser. II.
**94**(2), 151 (2016).View ArticleGoogle Scholar - E. Gordeev, O. Girina, E. Lupyan, A. Sorokin, L. Kramareva, V. Y. Efremov, A. Kashnitskii, I. Uvarov, M. Burtsev, I. Romanova, et al., The VolSatView information system for Monitoring the Volcanic Activity in Kamchatka and on the Kuril Islands. J. Volcanol. Seismol.
**10**(6), 382 (2016).View ArticleGoogle Scholar - S. Sawant, S. S. Durbha, A. Jagarlapudi, Interoperable agro-meteorological observation and analysis platform for precision agriculture: A case study in citrus crop water requirement estimation. Comput. Electron. Agric.
**138:**, 175 (2017).View ArticleGoogle Scholar - J. Padarian, B. Minasny, A. B. McBratney, Using Google’s cloud-based platform for digital soil mapping. Comput. Geosci.
**83:**, 80 (2015).View ArticleGoogle Scholar - K. Yumimoto, T. Nagao, M. Kikuchi, T. Sekiyama, H. Murakami, T. Tanaka, A. Ogi, H. Irie, P. Khatri, H. Okumura, et al., Aerosol data assimilation using data from Himawari-8, a next-generation geostationary meteorological satellite. Geophys. Res. Lett.
**43**(11), 5886 (2016).View ArticleGoogle Scholar - M. B. Hahn, A. J. Monaghan, M. H. Hayden, R. J. Eisen, M. J. Delorey, N. P. Lindsey, R. S. Nasci, M. Fischer, Meteorological conditions associated with increased incidence of West Nile virus disease in the United States, 2004–2012. Am. J. Trop. Med. Hyg.
**92**(5), 1013 (2015).View ArticleGoogle Scholar - S. Ansari, S. Del Greco, E. Kearns, O. Brown, S. Wilkins, M. Ramamurthy, J. Weber, R. May, J. Sundwall, J. Layton, et al., Unlocking the potential of NEXRAD data through NOAA’s Big Data Partnership. Bull. Am. Meteorol. Soc.
**99**(1), 189 (2018).View ArticleGoogle Scholar - J. C. Anjos, I. Carrera, W. Kolberg, A. L. Tibola, L. B. Arantes, C. R. Geyer, MRA++: Scheduling and data placement on MapReduce for heterogeneous environments. Futur. Gener. Comput. Syst.
**42:**, 22 (2015).View ArticleGoogle Scholar - B. Yu, J. Pan, in
*2015 IEEE Conference on Computer Communications (INFOCOM)*. Location-aware associated data placement for geo-distributed data-intensive applications (IEEE Computer SocietyWashington, DC, 2015), pp. 603–611.View ArticleGoogle Scholar - J. Paiva, P. Ruivo, P. Romano, L. Rodrigues, P A uto, lacer: Scalable Self-Tuning Data Placement in Distributed Key-Value Stores. ACM Trans. Auton. Adapt. Syst. (TAAS).
**9**(4), 19 (2015).Google Scholar - H. Chen, Y. Wang, A new CQ method for solving split feasibility problem. Appl. Math. Comput.
**218**(8), 4012 (2011).MathSciNetGoogle Scholar - H. Zhang, Y. Wang, A family of higher-order convergent iterative methods for computing the Moore–Penrose inverse. Front. Math. China.
**5**(1), 37 (2010).MathSciNetView ArticleGoogle Scholar - C. Hou, W. Yuan, The convergence of conjugate gradient method with nonmonotone line search. Math. Ann.
**353**(2), 499 (2012).MathSciNetView ArticleGoogle Scholar - G. Wang, H. t Che, H. b. Chen, Feasibility-solvbility theorems for generalized vector equilibrium problem in reflexive banach spaces. fixed point theory appl.
**2012**(1), 38 (2012).View ArticleGoogle Scholar - C. Hou, H. Zhang, Derivations of a class of Kadison–Singer algebras. Linear Algebra Appl.
**436**(7), 2406 (2012).MathSciNetView ArticleGoogle Scholar - Z. J. Shi, S. Wang, Z. Xu, Minimal generating reflexive lattices of projections in finite von Neumann algebras. Mathematische Annalen.
**353**(2), 499–517 (2012).MathSciNetView ArticleGoogle Scholar - C. Hou, Minimal generating reflexive lattices of projections in finite von Neumann algebras. Linear Algebra Appl.
**466:**, 241 (2015).MathSciNetView ArticleGoogle Scholar - L. Huan, B. Qu, J. g. Jiang, Merit functions for general mixed quasi-variational inequalities. J. Appl. Math. Comput.
**33**(1-2), 411 (2010).MathSciNetView ArticleGoogle Scholar - R. Jean-Luc, F. Gonon, L. Favre, E. L. Niederhäuser, in
*2018 International Conference on Smart Grid and Clean Energy Technologies (ICSGCE)*. Convergence of Multi-Criteria Optimization of a Building Energetic Resources by Genetic Algorithm (IEEE Computer SocietyWashington, DC, 2018), pp. 150–155.View ArticleGoogle Scholar - R. L. Kadri, F. F. Boctor, An efficient genetic algorithm to solve the resource-constrained project scheduling problem with transfer times: The single mode case. Eur. J. Oper. Res.
**265**(2), 454 (2018).MathSciNetView ArticleGoogle Scholar - D. Gong, J. Sun, Z. Miao, A set-based genetic algorithm for interval many-objective optimization problems. IEEE Trans. Evol. Comput.
**22**(1), 47 (2018).View ArticleGoogle Scholar