Security First: Evaluating Cloud Services for Data Protection and Compliance

Authors

• Rakesh Mittapally Tableau/Micro Strategy Architect and Administrator, AARP Washington DC, USA

Abstract

For businesses looking to take advantage of cloud service, choosing the best cloud service provider is essential. It is necessary to take into account variables like scalability, reliability, security, and cost-effectiveness. It is crucial to evaluate the provider's infrastructure, service level agreements, support choices, and interoperability with existing systems in order to make a well-informed choice that satisfies particular business demands.

The impact that cloud service selection has on enterprises and organizations is what gives the topic study significance. Making informed choices about cloud service providers can have several advantages, including cost reduction, increased scalability, greater data protection, and simpler business processes. Organizations can choose cloud service providers that meet their unique needs by performing research on the topic. This results in more effective resource allocation, more productivity, and better overall performance. Additionally, research in this field assists in identifying new trends, best practices, and potential hazards related to the use of cloud services, assisting in the creation of frameworks and strategies for informed decision-making

Weighted sum method is a decision-making technique that assigns weights to various criteria and calculates a weighted sum to determine the best alternative based on those criteria.

Productivity, Accuracy Complexity, Flexibility, Material, utilization, Quality, Operation cost. Process Sand castings, Investment casting, pressure die casting, gravity die casting, and additive manufacturing are some examples.

Efficiency, accuracy, complexity, adaptability, material use, quality, and operating costs. From the result quotes of the day got 1st rank and least by historical stocks quotes First rank got by quotes of the day in cloud service

Introduction

Because many applications, cloud services and web services are closely related. Infrastructure services provide virtual environments that may be accessed by online interfaces and monitoring tools, whereas platform services support the development of web services [16]. Specific web service discovery approaches have been used for choosing cloud services, especially SaaS services. However, because cloud services cannot be merely categorized as web services, applying web service selection methodologies to the cloud environment presents significant difficulties. Although cloud services and web services are comparable, the latter are more extensive and include more variables. SaaS, Platform as a Service Traditional web service selection methods have difficulties due to the complexity created by the special characteristics of the cloud, such as scalability, elasticity, and resource pooling. A variety of challenges must be overcome in order to choose and use cloud services effectively. These difficulties include taking into account service-level agreements, performance measurements, privacy and security issues, system interoperability, cost models, and provider dependability. These extra variables must be taken into account, and new tactics must be developed to address the unique needs and requirements of cloud services, in order to adapt existing web service selection procedures to the cloud environment.

Although cloud services and online services are closely related, it is crucial to understand that cloud services cannot only be categorized as web services. It is essential to address the new issues and take into account the wider scope and distinctive qualities of cloud services in order to properly adapt online service selection methodologies to the cloud. This necessitates the development of specialized approaches that take into account the various requirements and aspects involved in choosing a cloud service [1]. There are different kinds of cloud services accessible depending on factors like cost and technical feasibility. Public and private clouds are included in the first division, which is based on the accessibility of services. Private clouds are only available to the owner organization and its affiliates, whereas public clouds are open to everyone. Depending on the kind of cloud service being offered, the second classification is determined. There are four subcategories within it: IaaS stands for "Infrastructure as a Service," and it refers to the concept where cloud service providers supply virtualized computer resources such virtual machines, storage, and networking infrastructure.

Operating systems, applications, and data are in the control of the customer, and the underlying infrastructure is managed by the cloud provider. Platform as a Service (PaaS): With PaaS, cloud users access the provider's computer infrastructure. This encompasses instruments, programming dialects, and frameworks that let users create, launch, and administer applications without having to deal about infrastructure administration. Software as a Service (SaaS): SaaS refers to the subscription-based delivery of software programmer through the internet. The cloud provider handles the complete infrastructure, including software updates and maintenance, while users access the programmers using web browsers. While private clouds can only be accessed by the owner organization and its subsidiaries, public clouds are open to everyone. Based on the kind of service the cloud provides, the second classification is made.

The following subcategories are included in it: infrastructure as a service These variations could relate to the particular criteria taken into account, the decision-making models used, the deployment of fresh methodologies, or the usage of other algorithms or approaches. By highlighting these differences, we hope to add to the body of knowledge already available and offer insights into the developments and unmet needs in the area of choosing web and cloud services [4].Many different techniques have been employed by researchers to address the CSRS problem, according to a thorough analysis of the current literature. For instance, many authors have proposed selecting cloud services based on their dependability [20–23. Since trust is a highly individualized term, it is impossible to establish a precise quantitative measurement of it [24].

As a result, these uni-dimensional models may result in biased recommendations that don't seem to offer a practical solution. Practically, the ideal person to determine and priorities which element is favored above others and to what amount is a DM who is familiar with and aware of the Qo S requirements of an enterprise. Similar studies can be found in [25] and [26]. The buyer is still vulnerable to being duped by dishonest brokers because there is no process for assessing whether such a tip is genuine. These dishonest brokers might market products and services that benefit them more. Does a client have a viable alternative to CSRS in light of the aforementioned logic? arises. Some claim that "Everything as a Service," or XaaS, is the ideal way to describe cloud computing. [28].. Can CSRS be provided as a service? This issue is pertinent in light of the aforementioned ongoing research issues, potential client resistance to cloud brokerage models, and the concept of XaaS. To the best of our knowledge, the existing research has not at all taken into account this dimension. [5].The usual way for evaluating cloud services is comparing the performance variances between comparable cloud services.

Typically, the basis of such a comparison is the results of a predetermined set of benchmark tools [11] [10]. Since cloud services are heavily virtualized, benchmark methods for evaluating the performance of traditional calculations can be employed efficiently in cloud contexts. By integrating these bench-marking tools in accordance with cloud characteristics, numerous metrics (such as CPU performance, memory read/write and storage, service response time, and throughput) may be quantitatively tested. However, benchmark testing results frequently can not be a reliable indicator of how well a cloud service works for typical cloud consumers.This is because a limited number of tests would not be able to accurately mimic the broad range of real-world activities being carried out in the cloud, and since testing conditions are frequently different from those utilised by regular users in their daily work. These benchmark tests also typically act as spot-check examinations. Continuous testing is difficult because it may incur expenditures that are similar to or higher than those related to employing an actual cloud service.[6].

The problem of selecting a cloud service belongs to the class of multi-criteria selection problems, and the traits of typical MCDM problems as described by [31] and [32] are comparable to the problem of selecting a cloud service and support the concept of an MCDM-based cloud service It is crucial to compare cloud services based on the variation in their performance over time in order to develop a grading system for them. Surprisingly, earlier investigations [5] did not fully take into account this component.Comparing performance differences between similar services is the traditional method for assessing cloud services. A specified set of benchmark tools is often used for this, as they offer impartial measurements for performance assessment [11] [10].Decision-makers can learn more about the reliability and consistency of service performance over time by analysing and contrasting the performance fluctuations of various cloud services. When evaluating the value and applicability of cloud services for particular needs, this data is essential.Since cloud services are highly virtualized, bench-marking techniques can be effectively used in cloud contexts to assess the performance of conventional calculations. Numerous parameters (such as CPU performance, memory read/write and storage, service response time, and throughput) may be quantitatively examined by integrating these bench-marking tools in line with cloud features.

The effectiveness of a cloud service for normal cloud users, however, is not always reliably indicated by the results of benchmark testing. services. [7]. In contrast to other online services like buying a flight or hotel room, the selection process for cloud services is distinct. This is true because cloud services are unique in and of themselves. IaaS (Infrastructure-as-a-Service), PaaS (Platform-as-a-Service), and SaaS (Software-as-a-Service) are a few examples of provisioning techniques used to provide various cloud service types. Different selection criteria will apply to PaaS and IaaS services compared to SaaS services like Salesforce, Google Apps, and Microsoft Office 365. PaaS and IaaS services include AWS Elastic Beanstalk, Rack-space, and Cisco Meta-pod. Additionally, there are differences in the quality of service provided by various cloud providers in terms of aspects like price, usability, performance, security, and privacy. varied clients and service providers may have varied expectations on the level of these attributes, which might be subjectively understood.

The absence of a widely agreed common benchmark for evaluating the grade of cloud services makes it more difficult to address this issue customers are unable to fully rely on these approaches. User Preferences (UPs), for example, are one such area where there is confusion. which define how satisfied a user is with the service, depend on user requirements. This level of pleasure may change during the course of utilising the programmer. None of the techniques outlined in the most recent research gather this information and make better service suggestions to potential clients. Customers may change their choices when utilising a cloud service, for instance, depending on how well the service functions for them. It is crucial to collect altered preferences from customers both before and after they use the suggested service in order to better understand user satisfaction.

As shown in a prior study, patterns and trends can be found by examining the satisfaction levels of users with comparable service requirements. Other clients who have comparable service requirements may find this information useful because it helps them choose a cloud service with knowledge. It is simpler to decide which cloud service is best for particular needs by taking into account the experiences and contentment of similar users.

Material and methods

The term "process and castings" typically refers to the process of creating castings, which are solid metal objects created by pouring molten metal into a mould and allowing it to cool before it solidifies. This approach is often used by many companies, including those in the automotive, aerospace, building, and other industries. Gravity die casting: This method, which is also known as permanent mould casting or chill casting, creates solid metal objects using a reusable metal mould. It is referred to as "gravity" die casting because the molten metal is poured into the mould using the force of gravity. Investment casting: Complex and complicated shapes can be produced using investment casting, sometimes referred to as lost-wax casting.

A wax template or reproduction of the intended object is made, which is then covered in a ceramic shell. A hollow filled with molten metal remains after the wax has been melted out of the shell. The ceramic shell is removed after the metal hardens to show the finished cast metal item. The manufacturing method known as additive manufacturing, sometimes referred to as 3D printing, creates three-dimensional items layer by layer from a digital model or CAD (Computer-Aided Design) file. Instead of the more conventional subtract manufacturing techniques like cutting or drilling, it entails the additive deposition of material. Productivity: A statistic called productivity is employed to evaluate the effectiveness and output of a manufacturing process or workforce over a predetermined period of time. It calculates the ratio of output to input, which shows how efficiently labour, time, or other resources are used.

Higher levels of productivity imply better efficiency and effectiveness in the manufacturing process as more output is produced with the same quantity of input. Monitoring and raising productivity levels is essential for streamlining processes, increasing production levels, and reducing resource usage. Accuracy Complexity: Accuracy and complexity are two important aspects in manufacturing processes, including casting, die casting, and additive manufacturing. Flexibility: Flexibility in manufacturing refers to the ability of a production system to adapt and respond to changes in demand, product variations, or new requirements. It involves the agility and versatility of the manufacturing process to accommodate different product designs, volumes, and production schedules. Material: The ability of a manufacturing process and its tools to efficiently deal with a wide variety of materials in order to produce different products or components is referred to as material flexibility in manufacturing.

Weighted sum method (WSM): serve as illustrations for this trustworthy methodology. In the first example, the weighted sum approach is used to quickly estimate the Pareto surface, producing a mesh of Pareto front patches. New equality constraints on a piece-wise planar hyper-surface in the m-dimensional goal space are added to each patch to improve it further. The pseudo nadir point is connected to the expected Pareto optimum solutions by these limitations. The technique works well at finding solutions in non-convex areas and creating a well-distributed Pareto front mesh, which makes it easier to visualize the outcomes. [3] There is a weighted sum method offered. In the related multi-objective optimization problem, the locations generated by this method are all Pareto optimal. In the MATHEMATICA programming environment, a technique is provided for generating weighting coefficients wi > 0. The symbolic run-time transformation of the associated single-objective constrained issue's constraints and objective functions is the main point of interest. Implementation details and graphical depictions of the two- and three-variable examples are provided to demonstrate the suggested methodology. [4] Following regular weight adjustments, the weighted sum approach finds a distinct optimal solution for each and every objective optimization.

The generated solutions are nearly in ahead of the Pareto principle. The weighted sum method has the earliest study, according to Zadeh3. Koski4 employed the weighted sum method for structural optimization. Lin6 developed the equality constraint approach, whereas Marglin5 developed the -constraint strategy. Multi objective optimization was carried out using genetic algorithms, simulated annealing by Suppapitnarm7, and genetic algorithms by Goldberg8, Fonseca and Fleming9, among others. Multi-objective optimization also uses heuristic methods. This article's initial objective is to provide a modified version of the Weighted Sum Method (WSM) as a solution for decision-makers who must rank or weigh particular criteria but lack the necessary information. The proposed improvement is based on WSM, one of the most widely used Multiple Attribute Decision Making (MADM) methodologies.

The expanded WSM technique is presented in the text in five steps. An analysis of a numerical example that shows the updated WSM's practical applicability helps to validate the effectiveness of the updated WSM. The outcomes of this case demonstrate the validity and applicability of the enhanced WSM in decision-making contexts.6]It is known that the weighted sum approach of vector objective secularization produces locations on the convex Pareto front whose distribution is unpredictable. This study's nonlinear weight selection method can be used to enhance the weighted sum approach's distribution of Pareto points. Examples with numbers are given to show how effective the method is.7]A concurrent subspace optimization framework is described together with a weighted sum approach.

The adaptive weighted sum is employed in the bi-level concurrent subspace optimization framework to trade off several, competing goals. Two changes are performed to produce better distributed solutions. First, the bi-level optimization framework relaxes an additional equality requirement that causes sluggish convergence [20]. By using this normalization method, all criteria are checked against the. [71] employed the weighted product method (WPM) and aggregated WSM to determine the ideal wind power facility based on technical, economic, and environmental factors. is a commonly used and well-recognized technique., and practically applicable method for making subjective multi-criteria decisions that is simple to implement [13–16]. For practitioners with less mathematical background, WSM is advised among MADMs [17].

This strategy, according to Kumar and Suresh [18], entails decision-making procedures and techniques where each choice must be given a score based on a pertinent criterion, with each criterion being weighted according to importance. The methodical application comprises identifying the numerous levels of each criterion, assigning appropriate scores for each level, and determining the highest scores for each criterion. In order to provide a recommendation or rating that takes into account the range of qualities, the values, and the demands associated with a certain component, the process of shortlisting or screening the criteria necessitates merging and combining the data. [13].

Results and Discussion

Table 1. Data Set

Table 1 shows theAlternative:Throughput, Reliability, Portability, Response time, Evaluation preference: Quote of The Day, Xignite Quotes, Stock Quotes, Real Time Quotes, Delayed Stock Quotes, Historical Stock Quotes.

Figure 1.Data Set

Figure 1 showing that Alternative:Throughput, Reliability, Portability, Response time, Evaluation preference: Quote of The Day, Xignite Quotes, Stock Quotes, Real Time Quotes, Delayed Stock Quotes, Historical Stock Quotes.

Table 2. Normalized Data

Table 2 shows the Normalized Data for Alternative: Throughput, Reliability, Portability, Response time, Evaluation preference: Quote of The Day, Xignite Quotes, Stock Quotes, Real Time Quotes, Delayed Stock Quotes, Historical Stock Quotes. Residual error-it is also Maximum or Minimum value =C5/MAX ($C$4: $C$8), =MIN ($D$4: $D$8)/D6 Normalized Data formula used.

Table 3. Weight

Table 3 shows the Weightages used for the analysis. We take same weights for all the parameters for the analysis.

Table 4. Weighted normalized decision matrix

Table 4 shows the Weighted Normalized Decision Matrix. Alternative: Throughput, Reliability, Portability, Response time, Evaluation preference: Quote of The Day, Xignite Quotes, Stock Quotes, Real Time Quotes, Delayed Stock Quotes, Historical Stock Quotes.Residual error it is also Weighted Normalized Decision Matrix value multiplication formula used.

Figure 2. Weighted normalized decision matrix

Figure 2 shows the Weighted Normalized Decision Matrix. Alternative: Throughput, Reliability, Portability, Response time, Evaluation preference: Quote of The Day, Xignite Quotes, Stock Quotes, Real Time Quotes, Delayed Stock Quotes, Residual error it is also Weighted Normalized Decision Matrix value multiplication formula used.

Table 5.Preference Score & Rank

Table 5 shows the graphical view of the final result of this paper the Square is in 1^st rank, the Residual error is in 2^nd rank, the Quote of The Day is in 5^th rank, the Real Time Quotes is in 4^th rank, and the Xignite Quotes is in 3^rd rank. The final result is done by using the WSM method.

Figure 3.Preference Score

Figure 3. Preference Score shows the Stock Quotes 0.95778, Delayed Stock Quotes 0.94885, Xignite Quotes0.82968, Real Time Quotes 0.69531, Quote of The Day 0.69321.

Figure 4.Rank

Table 5 shows the graphical view of the final result of this paper the Stock Quotes is in 1^st rank, the Residual error is in 2^nd rank, the Regression is in 5^th rank, the Interaction is in 4^th rank, and the Linear is in 3^rd rank. The final result is done by using the WSM method.

Conclusion:

We concentrate on the service selection for cloud computing in cases requiring several criteria. We provide a taxonomic classification along with a description of the MCDA types and attributes. By examining and synthesizing the available literature, we contrast various approaches. There are several real-world instances of current applications of various techniques. Therefore, MCDA plays a significant role in situations involving several factors for decision-making. As a result, we list a number of the benefits and drawbacks, as well as numerous uses for these MCDA techniques in choosing cloud services. In addition, multi-criteria applied approaches are comprehensively summarized and collated for use in various study disciplines. Additionally, many MCDA methodologies and their distinctive characteristics are described and contrasted, which can help new researchers choose their research focuses.

References:

1. Sun, Le, Hai Dong, Farookh Khadeer Hussain, Omar Khadeer Hussain, and Elizabeth Chang. "Cloud service selection: State-of-the-art and future research directions." Journal of Network and Computer Applications 45 (2014): 134-150.
2. ur Rehman, Zia, Farookh K. Hussain, and Omar K. Hussain. "Towards multi-criteria cloud service selection." In 2011 fifth international conference on innovative mobile and internet services in ubiquitous computing, pp. 44-48. Ieee, 2011.
3. Sundareswaran, Smitha, Anna Squicciarini, and Dan Lin. "A brokerage-based approach for cloud service selection." In 2012 IEEE Fifth International Conference on Cloud Computing, pp. 558-565. IEEE, 2012.
4. Sun, Le, Hai Dong, Omar Khadeer Hussain, Farookh Khadeer Hussain, and Alex X. Liu. "A framework of cloud service selection with criteria interactions." Future Generation Computer Systems 94 (2019): 749-764.
5. Hussain, Abid, Jin Chun, and Maria Khan. "A novel framework towards viable cloud service selection as a service (cssaas) under a fuzzy environment." Future Generation Computer Systems 104 (2020): 74-91.
6. Qu, Lie, Yan Wang, and Mehmet A. Orgun. "Cloud service selection based on the aggregation of user feedback and quantitative performance assessment." In 2013 IEEE international conference on services computing, pp. 152-159. IEEE, 2013.
7. Rehman, Zia Ur, Omar Khadeer Hussain, and Farookh Khadeer Hussain. "Parallel cloud service selection and ranking based on QoS history." International Journal of Parallel Programming 42 (2014): 820-852.
8. Eisa, Mona, Muhammad Younas, Kashinath Basu, and Hong Zhu. "Trends and directions in cloud service selection." In 2016 IEEE symposium on service-oriented system engineering (SOSE), pp. 423-432. IEEE, 2016.
9. Lin, Dan, Anna Cinzia Squicciarini, Venkata Nagarjuna Dondapati, and Smitha Sundareswaran. "A cloud brokerage architecture for efficient cloud service selection." IEEE Transactions on Services Computing 12, no. 1 (2016): 144-157.
10. Zhang, Miranda, Rajiv Ranjan, Armin Haller, Dimitrios Georgakopoulos, and Peter Strazdins. "Investigating decision support techniques for automating cloud service selection." In 4th IEEE International Conference on Cloud Computing Technology and Science Proceedings, pp. 759-764. IEEE, 2012.
11. Ilieva, Galina, Tania Yankova, Vera Hadjieva, Rositsa Doneva, and George Totkov. "Cloud service selection as a fuzzy multi-criteria problem." TEM Journal 9, no. 2 (2020): 484.
12. Nawaz, Falak, Mehdi Rajabi Asadabadi, Naeem Khalid Janjua, Omar Khadeer Hussain, Elizabeth Chang, and Morteza Saberi. "An MCDM method for cloud service selection using a Markov chain and the best-worst method." Knowledge-Based Systems 159 (2018): 120-131.
13. A utility-based approach for customised cloud service selection
14. Marler, R. Timothy, and Jasbir S. Arora. "The weighted sum method for multi-objective optimization: new insights." Structural and multidisciplinary optimization 41 (2010): 853-862.
15. Kim, Il Yong, and Oliver L. De Weck. "Adaptive weighted-sum method for bi-objective optimization: Pareto front generation." Structural and multidisciplinary optimization 29 (2005): 149-158.
16. Kim, Il Yong, and O. L. De Weck. "Adaptive weighted sum method for multi-objective optimization: a new method for Pareto front generation." Structural and multidisciplinary optimization 31, no. 2 (2006): 105-116.
17. Kim, Il Yong, and O. L. De Weck. "Adaptive weighted sum method for multi-objective optimization: a new method for Pareto front generation." Structural and multidisciplinary optimization 31, no. 2 (2006): 105-116.
18. Hazelrigg, George A. "A note on the weighted sum method." Journal of Mechanical Design 141, no. 10 (2019).
19. Ryu, Jong-hyun, Sujin Kim, and Hong Wan. "Pareto front approximation with adaptive weighted sum method in multi-objective simulation optimization." In Proceedings of the 2009 Winter Simulation Conference (WSC), pp. 623-633. IEEE, 2009.
20. Weeraddana, Pradeep Chathuranga, Marian Codreanu, Matti Latva-aho, Anthony Ephremides, and Carlo Fischione. "Weighted sum-rate maximization in wireless networks: A review." Foundations and Trends® in Networking 6, no. 1–2 (2012): 1-163.
21. Kim, Il Yong, and Olivier de Weck. "Adaptive weighted sum method for multiobjective optimization." In 10th AIAA/ISSMO multidisciplinary analysis and optimization conference, p. 4322. 2004.
22. Saravanan, Vimala, M. Ramachandran, and Malarvizhi Mani. "Selection of Photovoltaic Devices Using Weighted Sum Method." Renewable and Nonrenewable Energy 1, no. 2 (2022): 67-73.
23. Sorooshian, Shahryar, and Yasaman Parsia. "Modified weighted sum method for decisions with altered sources of information." Mathematics and Statistics 7, no. 3 (2019): 57-60.
24. Jubril, Abimbola M. "A nonlinear weights selection in weighted sum for convex multi-objective optimization." Facta Universitatis 27, no. 3 (2012): 357-372.
25. Zhang, Ke-shi, Zhong-hua Han, Wei-ji Li, and Wen-ping Song. "Bi level adaptive weighted sum method for multidisciplinary multi-objective optimization." AIAA journal 46, no. 10 (2008): 2611-2622.
26. Jakob, Wilfried, and Christian Blume. "Pareto optimization or cascaded weighted sum: A comparison of concepts." Algorithms 7, no. 1 (2014): 166-185