Understanding and Analysing Resource Utilization, Costing Strategies and Pricing Models in Cloud Computing

This paper is primarily focused on understanding the basics of cloud computing economics – specific pricing models and general cost structure and compares between basic cost structures. It also examines 15 pricing models dependent on various bases and attempts to consolidate the different pricing methods used to guarantee maximum profitability to both the service provider and consumer. To avail the services of cloud computing from the cloud service provider, the organization has to incur individual costs. There are three types of costs: fixed, variable, and semi-variable costs. Each type is computed and analyzed. There are more than 30 pricing models though the pay-as-you-go model, subscription model, and pay-for-resources models are in demand. To consider and discuss the basics of cloud computing, emphasizing its economics and pricing models considering various aspects and costing components, using and estimating plans from various cloud Registering suppliers. The paper also considers identifying gaps, if any, present in the current pricing models that are highly preferred with other pricing models that are designed but not implemented due to practical conditions and fear of higher percentage of risk and error and to compare the basic prices provided by the cloud service providers.


Literature review
Cloud Computing was introduced, as it turned into a need to have higher storage capacity and when the associations were hesitant to have all information on-premises. Cloud computing offers different types of assistance and organization models to guarantee that the consumer's needs are fulfilled. Cloud likewise helps in guaranteeing that all the users' can get to it correctly and give total security [22]. Later, the pricing and costing strategies of the cloud were highlighted since it was essential to understand these to be able to ensure that the customers' interests are well protected. Further explained as the basis of various services and deployment and the limitations of cloud computing. Many research authors classify the pricing models and throw light upon static pricing, limiting Service provider's profits. Many authors explain the various circumstances for the usage of the pricing models and believe that the Cloud providers form an "Oligopoly." In [1] also mention the basic cost structure is followed and the factors that affect the price of cloud resources. It is believed that genetic pricing approaches must be explored and conclude that all static models are implemented. However, only a few dynamic models are implemented though they are fairer and more adequate. It is also mentioned that the service providers may be more benefited from the pricing schemes rather than the consumers. Various services are mentioned with the demand and supply of each in the market. The detailed, comprehensive study of various models is explicitly provided in various research materials. It is believed that the consumer is to choose an apt model as per their requirements and infrastructure and that it is the responsibility of the consumer to do it aptly.
In [2] provides a comparison and discusses several pricing models of Cloud Computing provided by individual service providers. Specific pricing models are explained in short descriptions with their features and area of implementation provided. The fluctuations concerning demand and supply are focused on, presently, for the pricing models. The pricing structure at Sales cloud is laid out. Many authors believe that the change in risk must be shared between service providers and the service consumer and that the models are to be more inclined towards the consumer's requirements. As advancement in cloud computing technologies and virtualization is seen, further research is conducted in the same area to understand and compute its costing.
It was proven that Cloud Computing helps to maximize profits with minimum output. There are explanations as to how the price and pricing affect both parties in the usage of cloud storage systems. Individual researchers analyzed and tried to explain why most approaches are "theoretical and not yet implemented in the real market and maybe favoring Cloud Service Providers." Many costing models to assess the cost of the administration from the specialist organization are proposed. There is a survey of different models and their groupings to comprehend the issues of different actualizing models. Different issues of embracing cloud are additionally distinguished, and different worries of estimating are tended to while recognizing the issues on particular unique evaluating model execution. It is further understood through vigorous analysis and experimentation that "A single efficient model being created or suggested is a herculean task; instead, a comprehensive framework is to be built." "A balance in the budget is suggested to be fair in terms of both the parties. Pricing Strategies are considered to be highly complex buildings since it involves economics. Optimum utilization of resources and allocation is to be ensured by both the parties".
It is at this crossroad that the semi-variable costing technique is reviewed, and "Detailed explanation on cloud computing services with costs, only costs directly related" is derived; there is also a discussion on "Yield Management, promoting demand, better utilization, and reduction in idle capital. The author of the paper explores energy storage capital, power cost fluctuations, and competition to cloud service providers. The author also explains the value of the market and analysis cloud resource prices. In the paper, uniform distribution theory and independent values deriving key insights from such distribution are considered."."A mathematical model is discussed for problems as various factors as ideal QoS and resources being directly proportional to cost and explains various factors directly influencing price and presented with regression models specifically providing data analysis for the same.". Many experiments are conducted, concluding that the experimentation model is booming since the model presented results in an improvement in pricing policy transparency, Concluding that the future aspect of dynamic pricing being more in demand and in-use due to financial metrics and advantages it provides. However, individual researchers concluded: "with a preference for fixed pricing concerning cloud service providers with example.".
The research in cloud computing continues with technological and cost developments. The "Gartner's Hype cycle is presented for explaining the Total Cost of Ownership (TCO) model for Cloud Computing, The TCO model's approach and representation with reliable decision support." Cost Accounting is co-related to the TCO model by "Cost-benefit analysis." The advantages of the TCO model and future work aspects are discussed by various mathematicians and cloud computing researchers to understand how far the benefit of the TCO approach is. However, the model is not yet implemented, only theoretical. One of the papers' reviews provides the approach and process of development with assumptions required and cost structure with cost types and descriptions. Various service models' pricing schemes are broken down, and a mathematical approach with cost correlation, formulas, and mathematical modeling is provided.
Another researcher "classifies Cloud Computing Models (CCM) into eight types," explaining the Cloud Cube Model and how each model fits into CCM. Each of the eight types is further explained with its advantages and disadvantages. It concludes with a "proposal to Financial Cloud Framework (FCF) and future expansions of CCM into various fields."."An SBIFT (Scope, Base, Influence, Formula, Temporal Right) model is designed, explained in detail, and a correlation to the cloud is conducted, proposing a 7-dimension model". A further breakdown of the cost structure is provided. "An Addition to the SBIFT model is shown as part of the proposal-the Degree of Discrimination and the Dynamic Pricing Strategy." ComparisonComparison of the models designed is provided, and patterns of popular models are also analyzed and explained.
There is a comprehensive study ongoing for grid computing and cloud computing pricing models for each organizational goal, and these are compared with one another, with an explanation of grid and its types with an analysis of how it works. Six pricing schemes used in the grid are analyzed for pricing and found the methodology of implementation. Basics of service provided on Cloud Computing and vital technologies are discussed in various forums' of cloud computing wherein a comparison between grid and cloud computing from various aspects are provided.  Figure 1 shows the Types of Costs Representation. The procedure of valuing can be ordered into two regular evaluating models, for example, a fixed pricing model and a dynamic pricing model. These are two significant models to choose the expense of inquiry: accessibility when a client requests "something from the cloud, a question is terminated to the cloud and time skyline, i.e., when demand an inquiry." The contrast between fixed and dynamic is that in fixed estimating, every asset type has a predefined value, set by the dealer while utilizing dynamic estimating, processes each solicitation as per the evaluating instrument utilized. Both fixed and dynamic evaluating have the favorable circumstances that portray them in their utilization. For example, while the fixed cost is all the more straightforward for the client, dynamic estimating is reasonable for it; while static evaluating bolsters protections, dynamic valuing underpins suppliers to amplify benefits with every consumer. Figure 2 shows the Types of Costs.

Figure 2: Types of Costs
Fixed pricing: It is additionally called static estimating or pricing because it reliably for a more drawn-out timeframe. The asset has a foreordained value set by the supplier. It bolsters more confirmation for clients. Better to implement on buyers proportioning risk. Produce costs that do not fluctuate in the capacity of the customer. More comfortable to comprehend and all the more candid for users. Supports protections, steadier, diminish risks. Makes benefit estimation simple. It prohibits changing the cost as for time and cost. It could not be reasonable for all consumers. Does not permit suppliers to change cost for any reason. Unfair for the supplier (customer may pay not as much as his/her genuine utilization). Consumers may be charged for assets they have not consumed. It could be halted if shoppers come to the maximal limit. Are not founded on ongoing economic situations. Figure  3 shows the Cost Components.  Dynamic pricing: Setting off the cost of the administrations is truly adaptable because costs are differing concerning time. The asset will mean each solicitation dependent on evaluating components utilized. It modifies the cost from the point of view of cost and time. Reasonable for purchasers. The firm defers its valuing choices until secondary selling uncovered its conditions. Supports suppliers to boost benefits with each consumer. Setting cost depends on the present status of the market. It drives estrangement in the client. It needs more development innovation for altering cost and benefit calculation. Consumers may follow through on more significant expense than can pay. Many firms sometimes fall short of costs to react to advertised conditions. Many clients are not keen on this model. Consumers who address more feel inequality. Enforces a cost chance on shoppers.
Variable costing technique empowers a gauge for the fixed expenses, and variable expenses can be found in a brief time frame, with just fundamental arithmetic and no costly projects to run the estimations, considering the firm to contribute their limited assets somewhere else. This is especially helpful for small and micro firms which do not have budgets or the spending plan to bear the cost of more qualified experts.
It tends to be effortlessly determined and does not need a specialist for estimation. Arranging and Control; Financial arranging expects directors to gauge future prerequisites, which can be determined easily. This thus permits simple dynamics. It likewise makes the valuing more straightforward for Services Providers. The cost can be adequately taken care of by both gatherings. This technique is viable since different expenses do not influence it; it is needed when required. However, the issue with such costing is that it requires the fixed expense to be independently determined. It is likewise not entirely suitable for suppliers to just have variable costing. Specific segments require fixed costing, wherein this strategy will not be advantageous. Sometimes variable costing might be pointlessly given a more extensive centrality than it merits. Figure 4 shows the Types of Costs and Behavioural patterns.

Figure 4: Types of Costs and Behavioural patterns
Semi-Variable Pricing: In this type of costing and pricing, a part is fixed for a slot or for some time or up to the maximum unit, after which the calculation is conducted for Per Unit or, in this case, Per GB/TB. A semi-variable cost, otherwise called a semi-fixed expense or a mixed expense, is a cost made out of a blend of both fixed and variable segments. Expenses are fixed for a set degree of creation or utilization and become variable after this creation level is surpassed. This empowers a gauge for the fixed expenses, and variable expenses can be found in a brief time frame, with just essential arithmetic and no costly projects to run the figuring, considering the firm to contribute their limited assets somewhere else. This is especially helpful for small and micro firms which may not have the required funds for the same. The disservice of figuring semi-variable expenses through this specific technique is that it would disparage the expense as it does not separate the fixed and variable costs, prompting the expansion in use being disregarded and bringing about off base estimates.

Methodology Tools and Techniques
The methodology followed in this paper is qualitative study and analysis in terms of basics of Cloud Computing and the economics involved in the same. Similarly, secondary sources were used for collecting data related to the pricing of various service providers to compare them and provide an overview of the pricing strategies involved.
Most of the data collected are secondary data and from research papers and various websites. The ComparisonComparison is based on the introductory pricing for Asia (south-Mumbai). The analysis is less comprehensive since the data is gathered from secondary sources and as only as much is available for conducting research. All the experiments are conducted from the data collected from the respective Service Provider's websites. There may be a varied opinion concerning the experiment since the data collected is General and not considering each organisations' requirements and Service Level Agreements (SLAs).Also, since the experiments were based on calculations understood from that provided on the website of the service providers, it is not a specific mathematical model but just necessary calculations of cost. All these are also carried out without taking into consideration that there may be more models that are being used by the providers.

Basic Cost computation
The Basic mathematical formula for calculating Total cost for data storage on cloud is: T = c + I + m + s (Ellman et al., 2018) Total cost = computing cost + ip address cost + data storage cost + support cost Wherein, T = total cost of the entire process Computing cost is the cost of the principal componentsIP address cost is for the IP address, especially if it is a reserved IP address which also is not used to 100%.Support costs are the extra cost which is provided for various services that the cloud providers provide, such as development, business strategy.Data storage cost is the cost per GB which could either be calculated on an hourly basis or monthly basis. The Real-time examples considered of GCP, AWS and Azure as per their general pricing models are considered every month.
The Various Cost components are Hardware, Software, People, Accommodation, External Services, Transfer / Migration, Initial cost / Investment, Lease period/contract time, Quality of service, Rate of depreciation, Age of resources, Cost of maintenance. Other factors that affect the costing and pricing structure in cloud storage are provider's reputation,Capital cost of data centres, User review, Monitoring services, Social category of customers, Service level agreement (SLA), type of co-cloud user.
The basic pricing strategies used in the three providers are considered, and the experiment is conducted, wherein different costing methods are used according to the costing, and the most costefficient model for the consumers are highlighted. The calculations and experiments are conducted by understanding the calculation provided by the providers on their websites, and then the type of costing is considered for completing the calculation. Break down of cost structure in GCP for a month in general. Table 2 shows the Break down of cost structure in GCP, source -Cloud Storage Pricing.

Amazon Web Services (AWS):
The pricing scheme for a specific activity of scalable cloud storage. Table 3 shows the Amazon Web Services (AWS) Cloud pricing scheme, source -Amazon S3 Pricing.

Results
The fixed cost (salary, purchase, maintenance) is maximum, and variable costs are equally high due to technological factor in-house servers. Nevertheless, when it is a cloud, the fixed costs reduce and there is a higher rate of variable and semi-variable costs though much less compared when it is in-house servers since the cost of the repair is also reduced. It works on the "pay as you go "concept mainly. It is also understood that the providers used the Slot system or tier system for categorising the requirement of their clients and the cost with each tier though fixed when seen from the provider's perspective, the costs are variable costs. Moreover, as per the slots (tiers) created. From various sources being analysed, it can be concluded that; between providers, costs can vary by 17%, 27% of servers, which are operated as reserved instances, consume the majority of the costs. Costs could be optimised by studying cycles of peak activity. Expenses related to personnel, power, training. Have not been considered for this paper. The various Cost Models are being computed to ensure maximum EFFICIENCY is attained with OPTIMUM UTILISATION OF RESOURCES. From the GCP Basic pricing structure, we understand that the dynamic pricing used by them is the most cost-efficient in the perspective of the consumer. Google Cloud Storage estimating depends on the accompanying segments are Data storage, Network usage, Operations usage and Retrieval and early deletion fees. Various activities, like using AI, AI, ML, cloud textto speech. The costs are calculated in a tier system through the buckets selected by the consumer. Their Tier system allows various associations with varied requirements being satisfied with provided dynamic pricing. When fixed costing is implemented, the price increases, similarly, even when semi-variable costing is implemented, the price increases.
The total cost reduces when dynamic/ Variable pricing method is adopted, but this may benefit only the consumer, for the benefit of both the parties Semi-variable costing may also be adopted. GCP