Microsoft Word - 001.docx


 CHEMICAL ENGINEERING TRANSACTIONS  
 

VOL. 66, 2018 

A publication of 

 

The Italian Association 
of Chemical Engineering 
Online at www.aidic.it/cet 

Guest Editors: Songying Zhao, Yougang Sun, Ye Zhou 
Copyright © 2018, AIDIC Servizi S.r.l. 

ISBN 978-88-95608-63-1; ISSN 2283-9216 

SLA-Based Flexibility Cost Strategy for Cloud Computing 

System Architecture in Chemical Industry 

Ming Jianga, Zhenyu Liub 

School of Management, Depart. of Management Science, Xiamen University, Xiamen 361000, China 

mingjiang616584@126.com 

Under multiple pressures of environmental protection and increasing in the cost of raw materials, the 

production and operation costs of chemical companies remain on the rise. It is of great sig-nificance to 

investigate how to reduce the production and operation costs of chemical companies, provided that the flexible 

demands of the basic systems in chemical companies are met. The application of the flexible cost strategy of 

cloud computing system offers an opportunity for improving the cost management efficiency of chemical 

companies. This paper adopts the Operational Research, aiming at the flexible cost of the cloud computing 

system architecture, to consider the SLA flexibility requirements and machine-enhanced learning feedback of 

the cloud computing system architecture in the flexible strategy decision model, discusses and compares what 

kind of strategies the chemical companies adopt in various situations, in an attempt to pro-vide reference for 

chemical companies to make appropriate strategies when developing cloud computing systems. 

1. Introduction 

Aiming at the SLA in cloud computing, many scholars have conducted studies on different di-mensions. Some 

proposed relevant models for cloud users and cloud providers to maintain the availability of SLA through 

negotiation process, e. g. Alhammed et al. put forward the credi-bility model for evaluating cloud services that 

could provide a new set of criteria for cloud pro-vider to select these; some focused on maximizing the 

benefits of cloud service providers or cloud providers by the service quality in SLAs, e. g. Alhammed et al. 

provided a SLA framework outlined to maintain the credibility and reliability of cloud users and providers; some 

scholars proposed an architecture that could improve cloud service SLA supervision capacity, e.g. Mario et al. 

introduced a method to maximi ize the benefits of cloud provider with an intermediate entity, i.e. the economic 

expansion resource management. When the cloud provider cannot satisfy the demands of the SLA, this 

intermediate entity will provide some SLA references in light of possible losses in the next step, regardless of 

reputation losses as these methods may lead to. 

The cloud computing system architecture is the software that provides cloud services. It connects up cloud 

components with a coupling mechanism. Because of the complexity of cloud computing system architecture 

itself, it is immature to analyze it based on technologies and models, given such elements as performance, 

safety and reliability. For this purpose, scholars and cloud providers have proposed a variety of different could 

computing system architectures for different purpose and from different angles. Some scholars regard the 

cloud computing system as a tool, and think that the cloud computing system architecture should describe its 

internal technical characteristics. Some believe that the architecture should far less restrict specific 

technologies, but concern commonality and focus on the integrated model. The cloud computing system 

architecture refers to a kind of model that can access shared resources well allocated on demand via the 

Internet whenever and wherever possible. Some scholars also focus on the presentation mode of cloud 

computing systems. It, as a cloud computing service, runs on cloud computing networks. Its architecture is 

designed based on demands for various cloud services, as a keystone this study focuses on. 

CCS flexibility is described as a new computing model that extends the flexibility of IT with IT services 

available everywhere. Chebrolu believed that CC-based deployment itself is a mode of IT flexibility. 

Virtualization is considered to be one of the major implementation technologies for cloud computing system 

architecture flexibility. By the virtualization flexibility, the original fixed CPU, memory, and broadband are 

                                

 
 

 

 
   

                                                  
DOI: 10.3303/CET1866167 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Please cite this article as: Jiang M., Liu Z., 2018, Sla-based flexibility cost strategy for cloud computing system architecture in chemical 
industry, Chemical Engineering Transactions, 66, 997-1002  DOI:10.3303/CET1866167   

997



configured into multiple virtual machines to facilitate the horizontal expansion of underlying resources in the 

cloud computing system. while the modularity of the platform and application layers achieves the overall 

flexibility of the entire cloud computing system architecture. 

Based on the review of the above literature, the flexible cost of the CCS architecture is defined here as the 

sum of overall cost paid by company for appropriate flexibility in the entire life cycle of CCS architecture to 

meet the demand of user SLA in the cloud. 

2. Literature review 

With the rapid change of the social economy and the convenience of entering the Internet era, the business 

model and environment have changed greatly. On the one hand, enterprises need to respond quickly to 

demand, absorb, form and reconstruct the business process according to the fast changing environment, and 

continue to adopt innovative actions to cope with the fierce market competition. On the other hand, the 

generation of large amounts of data requires a corresponding technology and timely processing to form 

competitiveness. Cloud computing appeared in this background in the business field in 2006. It is the 

representative of the third IT technology after the personal computer and the Internet, and also the 

development trend of the next generation of IT technology architecture. It has become the most influential 

model in IT industry because of its advantages such as resource virtualization, and it is gradually applied to 

many fields, including business, medical and health, financial services, information technology, and even 

known as the fifth basic element except for water, air, electricity and telephone. At the same time, the 

development of the adaptive system, the monitoring module as the necessary module of the cloud computing 

system, has a small impact on the flexible cost of the architecture of original cloud computing system. The 

wide application of DevOps tools makes the cost of the cloud computing system has a great change 

compared to the original cost of the traditional information system architecture during the operation of the 

traditional information system architecture. It has become a hot topic how to effectively meet the SLA needs of 

cloud users and effectively control the flexibility cost of cloud computing system. Flexibility is an indicator of 

competition between enterprises and supply chains, which are equal to cost, quality and delivery time. It is the 

ability to respond quickly and follow changes when the environment, conditions, or process states change. 

Cloud computing system architecture is a software system architecture that provides cloud services, and the 

cloud components are connected by coupling mechanism. Because the architecture of the cloud computing 

system itself is very complex, it needs to consider such elements as performance, security, and reliability. It is 

not mature to analyze enterprise cloud computing system architecture by technology and model. Therefore, 

according to different needs and perspectives, scholars and cloud providers have proposed many different 

cloud computing system architectures. Some scholars regard the cloud computing system as a tool, and think 

that the architecture of cloud computing system should explain its internal technical characteristics. Some 

scholars believe that the architecture should be less limited to specific technology, but needs to care for 

generality and overall model. The cloud computing system architecture refers to a pattern that can be 

accessed by the network at anytime and anywhere. Some scholars also focus on the expression of cloud 

computing systems. Cloud computing systems are cloud computing services running on cloud computing 

networks. The cloud computing system architecture is designed according to various cloud service 

requirements. 

The flexibility of cloud computing system is described as a new computing mode, which expands the flexibility 

of IT through the available IT services. Vilalta and other scholars believe that deploying cloud computing is a 

way of IT flexibility. Virtualization is considered to be one of the main technologies of cloud computing system 

architecture flexibility. Through virtualization flexibility, the original fixed CPU, memory and broadband are 

configured into multiple virtual machines to facilitate the horizontal expansion of the underlying resources of 

the cloud computing system. At the same time, the overall flexibility of whole cloud computing system 

architecture is realized through the modularity of the platform layer and the application layer (Vilalta et al., 

2017). Hashem and Ibrahim believe that cloud computing is not only the application of services through the 

Internet, but also system resources and hardware resources that provide data centers for these application 

services (Hashem and Ibrahim, 2015). Cavalcante and Everton believe that cloud computing is a combination 

of high level platform resources and low level infrastructure resources, and users' access to cloud services is 

achieved on demand and paid in the way (Cavalcante and Everton, 2016). Cuenca-Alba and Jesús have a 

more comprehensive definition of cloud computing: cloud computing is a large-scale distributed computing 

paradigm driven by economies of scale, a large number of abstract, virtualized, dynamic, and manageable 

computing resources, storage capabilities, platforms and services across the Internet for external users to be 

distributed (Cuenca-Alba and Jesús, 2017). Ren and Lei believe that cloud computing is provided to 

consumers in the form of a variety of computing resources, storage resources, platforms and applications 

through services, and the consumers of the service need not know how the resources they use are provided. 

998



In the cloud environment, the cloud should not only meet the needs of the users, but also reduce the cost of 

the cloud computing, and improve the resource utilization ratio (Ren and Lei, 2017). Bobek and other 

researchers believe that flexibility refers to the ability of the system to deal with changing environment or 

environmental uncertainty (Bobek et al., 2017). Bai and other researchers believe that flexibility is the ability of 

the system to handle changes. This capability needs to be analyzed from the perspective of the third nature 

and stability. And they also identified a number of ways to test flexibility, such as, to assess flexibility by 

detecting the economic consequences of a system adapting to changes (Bai et al., 2017). Wang and Fei 

define flexibility as a system's ability to respond effectively to changing environments. Two types of flexibility 

are described: behavioral flexibility and state flexibility. The former refers to the ability to adapt to change 

through the interference from the outside of the system, that is, to increase the ability of the system to adapt to 

change through the structural changes in the organization and process. In contrast, the latter refers to the 

ability of the system to operate well in a different environment, that is, the system is not affected by the 

changes in the environment, but it ensures the relatively stable and effective operation of the production 

system (Wang and Fei, 2017). 

To sum up, in the above research work, the cloud computing system is compared with the original system, 

and one of the most important features is flexibility. While meeting the SLA requirements of the cloud users, 

the flexible cost of the cloud computing system needs to be effectively controlled. At the same time, the 

instability of the social and economic environment, the increasing uncertainty of the market, and the enormous 

pressure of the enterprises, all require enterprises to reduce the cost and improve the efficiency by adopting 

new technology. If the cost and output of the cloud computing system cannot be well controlled, the 

enterprises may not be able to benefit from them. Because of the constraints of resources, few related 

research guides them to control the cost and achieve sustainable development in the case of meeting the 

flexible needs of the cloud users. Therefore, based on the above research status, the method of operational 

research is adopted to subsection the flexible cost calculation formula of each stage, which provides the basis 

for enterprise decision making, and enables the enterprise make the appropriate flexible cost strategy to 

achieve the optimization of efficiency. 

3. Decision model for flexible cost 

This model is proposed by Schober and Gebauer based on the basic investment cost model for information 

systems, given the impact of the SLA flexibility requirements and machine-enhanced learning feedback of 

CCS architecture. The flexible cost decision process is shown in Figure 1 below. 

 

Figure 1: Schematic diagram of flexible cost strategy process 

999



The flexible cost decision process consists of three phases. Phase 1 is the design phase. The decision is 

made based on whether a static flexible strategy or a dynamic flexible strategy is adopted. This flexible cost in 

this period mainly includes basic components and adaptive costs for supporting static flexibility, as well as 

dynamic costs for supporting dynamic evolution. Phase 2 is the operation and maintenance (O&M) phase, 

including three types of flexible strategies, i.e. static, dynamic, manual flexible strategies, which correspond to 

the O&M, upgrade, and manual costs, respectively. The third stage includes manual and external flexible 

strategies, which correspond to manual and external costs, respectively. 

The total cost, TCOST, for the cloud computing system architecture in Figure 1 includes the following several 

parts: ICOST is the cost when the company adopts the static flexible strategy at the design time; FCOST is 

the cost when the dynamic flexible strategy at the design time; UCOST is the cost when company upgrades 

the dynamic flexible part originally designed for CCS at the time of the operation and maintenance; OCOS is 

the cost when company upgrades the static flexible part originally designed for CCS at the time of the O&M; 

MCOST is the cost when company manually repairs external flexible strategy adopted within the time as SLA 

requires at the time of the O&M; PCOST is the cost when company can not repair the external flexible strategy 

manually within the time as SLA requires, but by other method at the time of O&M. At this time, the enterprise 

adopts an external flexible strategy and can manually repair the costs within the required time of the SLA. 

When the PCOST is operated and maintained, the enterprise adopts an external flexible strategy, which 

cannot be manually repaired within the required time of the SLA, but through other methods. 

TCOST = ICOST + FCOST + OCOST + TMCOST + UCOST (1) 

The following formulae are explained: 

3.1 ICOST is the cost of the static flexibility strategy that the company adopts at the design time: 

ICOST = {a + bL(𝑥1)[q + (1 − q)DC]}z (2) 

a is the fixed cost; 

b is the variable cost; 

q is the ratio of completed development until now; 

L(𝑥1) = 𝑥1
𝑣 [1 − (1 − 𝑥1)

1−𝑣 ] Lorenz coefficient curve, used to represent the ratio of tasks that have been 

processed. The greater v is, the more the concentration of tasks is, that is, the greater the difference is; 

DC =
((1 + i)𝑇 − 1)

(𝑖(1 + 𝑖)𝑇 𝑇)
⁄  represents the average discount rate; i is the annual average interest rate; T 

is the life cycle of the CCS; z is a binary number, used to ensure the availability of the calculation formula, 

then the following conditions are true: [1]  z ∈ (0，1)；[2]𝑥1 + 𝑥2 > 0，then z = 1; otherwise z = 0；[3] z ≥

0.5(𝑥1 + 𝑥2)； 

3.2 FCOST is the cost when company adopts the dynamic flexible strategy at the design time: 

FCOST = cy (3) 

c is the cost of dynamic flexible strategy;  

y is a binary number, used to ensure the availability of the calculation formula, then the following conditions 

are true [1] y ∈ (0，1); [2] in the design phase, when company adopts dynamic flexible strategy, y = 1 and y ≥

𝑥2； 

3.3 UCOST is the cost when company upgrades the dynamic flexible part originally designed for CCS 
at the time of O&M: 

UCOST = eL(𝑥2)DC (4) 

e is the cost that company performs this operation;  

  L(𝑥1) = 𝑥1
𝑣 [1 − (1 − 𝑥1)

1−𝑣 ]Lorenz coefficient curve, used to represent the ratio of tasks that have been 

processed. The greater v is, the more the concentration of tasks is, that is, the greater the difference is; 

DC =
((1 + i)𝑇 − 1)

(𝑖(1 + 𝑖)𝑇 𝑇)
⁄  represents the average discount rate; i is the annual mean interest rate; T is 

the life cycle of the CCS; 

There is a relationship between static, dynamic, external flexibilities and uncertainty. Due to the adoption of 

monitoring components, adaptive will continuously improve the uncertainty over time training, so that the 

uncertainty here refers to a function of time t, the specific strategies are as follows: 

1000



𝑤1 = 𝑝(𝑡)𝑥1; (5) 

Dynamic flexible strategy: 

𝑤2 = (1 − 𝑝(𝑡))𝑥2; (6) 

External flexible strategy: 

𝑤3 = 1 − 𝑤1 − 𝑤2 (7) 

3.4 OCOST is the cost when company operates and maintains the upgraded parts based on the static 
flexible and dynamic flexible strategies originally designed for CCS 

OCOST = Sd{(q + (1 − q)𝐿−1(0.6))w1 + 0.4w2}TDC (8) 

d is the O&M cost of parts based on the static and dynamic flexible strategies originally designed for CCS;    

S is the scalability of business loads during O&M; 

T, DC, w1, w2, q, and L(x) are given as above. Since the CCS is also an information system, 60% tasks 

proposed by Mutschler [31] are finished by the static flexible strategy, while 40% done by a dynamic flexible 

strategy 

3.5 TMCOST is the cost when the chemical company adopts the external flexible strategy during O&M. 
It includes the costs that company can manually repair it within the time as specified by SLA and 
exceeds the SLA time limits, only the compensation costs need to be met in other ways: 

TCOST = MCOST + PCOST； (9) 

When MCOST is the cost when the chemical company adopts an external flexible strategy that can be 

manually repaired within the time required by the SLA: 

MCOST = (hSf(1 + rg){((1 − q)(1 − 𝐿−1(0.6)))w1 + 0.4w2 + w3})TDC； (10) 

h is a binary number, which takes 0 when the SLA request time is exceeded, otherwise it takes 1; 

f is the cost of manual repair for the cloud computing system within the time required by the SLA;  

r is the urgency of the repair time; 

g is the cost-plus rate of time urgency; 

T, DC, w1, w2, w3, q, L(x), S are the same as above; 

When PCOST is the cost when the company will be unable to repair the parts of external flexible strategy 

within the time as required by SLA but by other method at the time of O&M;  

PCOST = ((1 − h)Sf(1 + rg){((1 − q)(1 − 𝐿−1(0.6))})w1 + 0.4w2 + w3})T𝑆𝐿𝐴 DC + (1 − h)K (11) 

TSLA is the time requirement for how long time the cloud user in the SLA takes to respond to the cloud 

provider;  

K is the cost that the cloud provider cannot repair the system within the time SLA requires. 

4. Discussion of strategies 

The flexible cost model of the cloud computing system architecture proposed in this paper involves three 

phases. In the design phases, it mainly involves the static and dynamic flexible strategy costs. The flexible 

cost of the static flexible strategy will also impact the cost in later O&M stage; dynamic flexible strategies can 

affect later upgrade and manual costs. Therefore, chemical companies should make decisions based on the 

flexible requirements of the SLA and the later O&M costs in the design phase, and a hybrid strategy has 

achieved the optimal effect. In the O&M phase, the major flexible costs include the O&M, upgrade, and 

manual costs. The chemical companies should make adjustments as required by the SLAs in time to avoid the 

external costs incurred in third phase. In the third phase, the chemical company should determine whether 

they continuously meet the requirements of SLA manually by negotiation or by external means such as 

compensation. 

5. Conclusion 

In relation to the original system, the cloud computing system features most the flexibility. How to effectively 

control the flexible cost of the cloud computing system on the premise that cloud user SLA in the chemical 

industries is satisfied becomes a hot topic. Simultaneously, the instability of the social and economic 

1001



environment, the increasing uncertainty of the market, the high pressure of environmental protection, the 

fluctuation of the costs of raw materials, and the huge pressure that chemical companies suffer from, all of 

these requires chemical companies to adopt new technologies to save costs and improve efficiency while 

satisfying the basic demands. Chemical companies may have not benefited from cloud computing system if 

they failed to well control its costs and output. Due to the constraints of resources, there are few studies that 

can provide the clues to the challenge on how to control costs and achieve sustainable development while 

satisfying the flexible demands of cloud users. the OR method adopted herein gives the flexible cost formulas 

at each stage to provide a basis for chemical companies to develop the future strategies, so that they can 

correctly make flexible cost strategies to maximize the efficiency. But badly, actual data inspection can not be 

performed, it needs to be supplemented in the later period. 

Reference  

Bai C.G., Joseph S., 2017, Improving green flexibility through advanced manufacturing technology investment: 

Modeling the decision process, International Journal of Production Economics, 188, 86-104, DOI: 

10.1016/j.ijpe.2017.03.013 

Bobek S., Grzegorz J.N., 2017, Uncertain context data management in dynamic mobile environments, Future 

Generation Computer Systems, 66, 110-124, DOI: 10.1016/j.future.2016.06.007 

Cavalcante E., 2016, On the interplay of Internet of Things and Cloud Computing: A systematic mapping 

study, Computer Communications, 89, 17-33, DOI: 10.1016/j.comcom.2016.03.012 

Cuenca-Alba J., 2017, ScipionCloud: An integrative and interactive gateway for large scale cryo electron 

microscopy image processing on commercial and academic clouds, Journal of structural biology, 200(1), : 

20-27, DOI: 10.1016/j.jsb.2017.06.004 

Hashem I.A.T., 2015, The rise of “big data” on cloud computing: Review and open research issues, 

Information Systems, 47, 98-115, DOI: 10.1016/j.is.2014.07.006 

Ren L., 2017, Cloud manufacturing: key characteristics and applications, International Journal of Computer 

Integrated Manufacturing, 30(6), 501-515, DOI: 10.1080/0951192x.2014.902105 

Vilalta R., Lopez V., Giorgetti A., 2017, TelcoFog: A Unified Flexible Fog and Cloud Computing Architecture 

for 5G Networks, IEEE Communications Magazine, 55(8), 36-43, DOI: 10.1109/mcom.2017.1600838 

Wang F., 2017, The values of market-based demand response on improving power system reliability under 

extreme circumstances, Applied Energy, 193, 220-231, DOI: 10.1016/j.apenergy.2017.01.103 

1002