Cloud Utility Price Models

Sururah A. Bello

and Christoph Reich

Obafemi Awolowo University, Ile-Ife, Nigeria

University of Applied Science Furtwangen (HFU), Furtwangen im Schwarzwald, Germany

Keywords:

Cloud Price Models, Utility Price Models, IaaS, PaaS, SaaS, CloudIA.

Abstract:

Cloud Computing’s service models have a number of proprietary pricing models as the services has been

commoditized to some extent. The SaaS especially so far has been known with a ﬂat price within the usage

time. Pricing models need to be more ﬂexible to prevent customers from thinking that paying same price for

a service over a period is no more cost effective, in spite of the level of utilization of the service. For Cloud

Computing this has to be taken into account. Africans have a business model with peculiarity. In order to

expand the acceptability of Cloud Computing to the African market, this peculiarity must be accommodated.

This study proposes cloud utility price models to give the Cloud customer the luxury of different usage style

and determine a customer speciﬁc, individual, most suitable price model. It gives the customer the opportunity

to choose a price model for the predicted usage and work within the budget.

1 INTRODUCTION

Cloud Computing has a pay-per-use business model

that is remarkably different from previous computing

paradigms. There is need for a means to transform the

computing services into monetary entities, thus the

need for an appropriate Business Model. Typically in

Infrastructure as a Service (IaaS) the commodities are

CPU power, memory, etc. for each virtual machine

(VM). For the Platform as a Service (PaaS) in gen-

eral the provider is commoditizing the availability per

month, used storage per month and used bandwidth.

PaaS is considered to be value added virtual instances

and is offered bundled together with the infrastruc-

ture. For the Software as a Service (SaaS) usage time

per month is mostly quantiﬁed. This commoditiza-

tion technique does not really portray the true utility

model of SaaS. In the existing models, once a VM is

started, the billing starts irrespective of the actual real

usage of a Cloud service. It should be distinguished

between a heavily used VM and an idle VM. This

study proposes an enhanced billing system that incor-

porates some minute entities for the SaaS. Differential

pricing resulted from quantifying this minute entities

will portray cloud further as utility service. The paper

is organized as follows: After the related work (Sec-

tion 2) various price models are analysed (Section 3).

Section 4 presented the taxonomy of usage items for

monitoring in Cloud services. Section 5 presents re-

warding models and personalized models. Section 6

shows the technical integration and Section 7 the con-

clusion.

2 RELATED WORK

Models are tools for expressing business logic and de-

scribing customers values (Osterwalder et al., 1998).

It is difﬁcult to apply the traditional system centric

allocation policy in a highly dynamic and distributed

environment. Hence the traditional business model

cannot be applied in the Cloud System. Transition

from the traditional Software Packaged delivery to

Software Delivery as a service in Cloud Computing

demands a new Business Model, as illustrated with

a Gaming Software ﬁrm in (Ojala and Tyrvainen,

2011). In (Chang et al., 2010), Cloud Computing

Business Models were classiﬁed into eight types and

related to a Hexagon Model, which illustrates the

sustainability of an organization through the adop-

tion of the right business model. Our study focuses

on the pricing models. The commercial success of

Cloud computing strongly rests on a cordial busi-

ness relationship between the provider and the con-

sumer through a good pricing system. In (Teng and

Magoules, 2010), an Auctioning Resource Pricing

policy allows users to predict the future resource pric-

317

A. Bello S. and Reich C..

Cloud Utility Price Models.

DOI: 10.5220/0004350503170320

In Proceedings of the 3rd International Conference on Cloud Computing and Services Science (CLOSER-2013), pages 317-320

ISBN: 978-989-8565-52-5

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

ing as well as satisfy budget and deadline constraints

without knowing offers from other users to solve the

price prediction problem. The study favors the cloud

provider. In (Ibrahim et al., 2011) it was established

that the prices charged for computation on IaaS is per-

sonally and socially unfair. The cost does not ac-

count for actual usage and charging is not uniform

the users. Hence proposed a new charging scheme for

IaaS, where users pay for actual consumption.

3 PRICE MODELS

According to (Rappa, 2004) water, electricity and

public transportation employed the metered usage

of service business models. While radio, TV, tele-

phone and Internet access uses mainly the subscrip-

tion model with some metered services for special

services. Cloud Computing is often seen as a part

of larger development towards long-dreamed vision

of society where computing is delivered as a utility

(Zhang et al., 2010). The cost of compute power

can be calculated by amortizing the capital cost over

the lifetime of the system as done in costing elec-

tricity systems. The total cost of a cloud service

will be a function of capital funds, depreciation, in-

terest rates and system lifetime. In (Jtmaa, 2010)

seven categories of cost in a Cloud System are iden-

tiﬁed, they are: server hardware, software, mainte-

nance, network, facilities, cooling and real estate cost.

These cost will be translated to the consumer in two

basic categories ﬁxed cost and the variable metered

cost. Fixed cost include, startup cost and availability

cost, this is synonymous with the access cost. The

variable cost is the operational cost, to be determined

by the consumption style. According to (Strømmen-

Bakhtiar and Razavi, 2011) the metered usage model

is applicable to products whose standardized qual-

ity is to some extent regulated. Therefore proposed

the metering model for IaaS (processing power, stor-

age devices, servers, I/O, other hardware etc). The

IaaS costing to include legal cost (e.g court or lit-

igation cost), data restoration and disaster recovery

cost (though a limit can be set) and regulatory re-

quirements (location dependent regulations, regular

backups). The metered model of the IaaS can have

variants like metering plus subscription or ﬁxed ba-

sic plus metered price model to allow for ﬂexibility.

PaaS is usually bundled with IaaS, so the PaaS cost is

incorporated as added value in IaaS. In costing PaaS,

customization can be allowed like a free start, billing

above free limit and metering price model. The sub-

scription and pay-as-you go model is proposed for the

SaaS layer. The subscription can also have consider-

ations for booking in advance and discounts for pre-

payment. Next is an overview of possible price mod-

els:

Static Price Models. Static Pricing is ﬁxed for a pe-

riod of time, and has a number of ﬂavors.

• Flat fee: Customers pay a ﬂat fee for a service for

a ﬁxed price.

• Menu Price: The customer pays a price that is al-

ready found on the catalog.

Variable by Market Value Price Models. On the

other extreme of the static pricing is a pricing mecha-

nism dictated purely by the market value. This pricing

mechanism has a number of variants.

• Haggling or Bargaining: The buyer and the seller

dispute the price to ﬁnalize on a productive agree-

ment.

• Yield Management: A pricing policy that allows

for optimizing proﬁt by anticipating, inﬂuencing

and forecasting customer behavior (e.g. hotel

rooms).

• Auctioning: This offer services for competitive

and open bid can either be Forward or Reverse

Auction.

• Dynamic Market: This ﬂavor results from contin-

ual change in both supply and demand of a prod-

uct or service. Only allows both buyers and sellers

to collectively change price.

Variable by Customer Characteristics Price Mod-

els. In another category of pricing mechanism, pric-

ing can be determined by the characteristics of the

customers, like the volume consumed and other cus-

tomer preferences but will not be affected by the mar-

ket value. This is also implemented in a number of

ways.

• Service Bundles: Prices can be set according to

the bundling of services features.

• Customer CV: The history of customers patronage

can also inﬂuence the price for a set of customers.

• Purchased volumes can as well be used to differ-

entiate prices for different customers.

The Customers valuation can also determine the ﬁnal

price of a product or service.

4 TAXONOMY USAGE ITEMS

Clouds have a pay-per use model, which implies the

usage monitoring of Cloud resources in clouds essen-

CLOSER2013-3rdInternationalConferenceonCloudComputingandServicesScience

318

tial. Based on a survey of the monitoring items of

e.g. Amazon, GO Grid, RightScale, and Salesforce, a

taxonomy of Cloud Use Items has been developed at

the level of IaaS, PaaS, and SaaS. From the provider

point of view, everything that causes cost is of inter-

est. From the customer point of view business related

values are more important.

Use Items in IaaS. Use items in IaaS are catego-

rized in Compute, Storage, Network resources with

their attributes (see Table 1). A long way to go is the

integration of QoS into the Cloud infrastructure, but

has to come, if Clouds want to be successful in busi-

ness areas.

Table 1: Use Items in IaaS.

Compute Resource

Attributes Explanation

No of Instances number of on demand

Type CPU performance (e.g. MIPS),

Memory size, etc.

Time of usage Typically per hour

OS Licensing issue

QoS Snapshot, backup, etc.

Storage Resource

Attributes Explanation

Storage Vol. Volume per month

QoS Redundancy, backup, etc.

Network Resource

Attributes Explanation

Msg transfer Upload, download, etc.

Components Need for ﬁrewall, routers

Use Items in PaaS. Table II lists the attributes in

the service model PaaS, which are not so wide spread

as the IaaS ones. Of course all the use items of the

IaaS can be taken into account at the PaaS level, but

there are additional ones.

Table 2: Use Items in PaaS.

Attributes Explanation

Deployment Data size, time of the deployment

Database Volume size, performance issues

Scalability Min/max limits, speed, etc.

Application Support for development,

development debugging, library support, etc.

Programming Type of programming language

language and possible licensing issues

QoS Availability., API supp., security

Use Items in SaaS. At present for the SaaS, only

the time of usage is being monitored. This study in-

tends to extend the monitored entities to items de-

scribed in Table 3.

Table 3: Use Items in PaaS

Attributes Explanation

No of user access More users cause more load

No of transactions Completed transactions

Usage history rewarding if experienced

customers

Usage period Daytime or night usage

QoS Request Response Time, etc.

5 REWARDING MODELS

Exchanging ownership requires methods of setting

worth, which is pricing. There is the need to com-

moditize to a standard level in order to price effec-

tively. Next rewarding models are discussed: Re-

warding Loyal Customers: Customer usage history

could also be used to initiate a price differential. To

retain existing customers, long usage history may be

rewarded. This may entice a customer to stay with

an old client. Economically, it is cheaper, since e.g.

an experienced customer requires less help and most

likely has fewer problems to deal with. Rewarding

Off Peak Usage: Usage outside off peak period could

initiate price differential. Instead of zero utilization

of SaaS services in some period, a personalized price

for such period might entice some SME users. E.g.

discount if the service is used between 10pm and 4am

in the morning. Rewarding Completed Transactions:

Service provider could give discount for a service that

generates more income (e.g completed transactions in

an online shop). Rewarding Increased Users: Moni-

toring the number of users accessing a service implies

service speciﬁc inspection. It might be used to adjust

the charge for the service. E.g. a very active service

may have it charge slightly raised to increase revenue

for the provider.

5.1 Personalised Model

Each of the above pricing introduces better oppor-

tunity for negotiation, thereby introducing ﬂexibility

which results in personalize pricing to the Cloud con-

sumer. The user is expected to adhere to the earlier

speciﬁcations, during the usage the actual usage style

is computed by the system to choose the best pric-

ing model. Business models that vary with customer

characteristics are known to be thriving in Africa. A

typical African is very comfortable with discrimina-

tory pricing. The customer has the feeling of cap-

turing surplus while the seller is also satisﬁed that a

desperate buyer pays more. Customers characteristics

are a good means of haggling. Africans are not used

to static pricing model. In order to expand the accept-

CloudUtilityPriceModels

319

ability of Cloud Computing to the African market the

variable pricing system need be incorporated deeply

into Cloud Computing. African buyers enjoy a feel of

personalize pricing.

6 INTEGRATION INTO CloudIA

CloudIA (Sulistio et al., 2009) is a private cloud es-

tablished by HFU to harness the potentials of Cloud

Computing for its internal usage and also for Small

and Medium Enterprises (SMEs). The CloudIA ar-

chitecture is based on OpenNebula with extra mod-

ules for monitoring, QoS, etc., and special security

issues. It has layers: the Business, the System and the

Resource layers. The proposed Utility Price Model

is incorporated into the Business Layer of CloudIA.

As shown in Fig. 1 a customer can access the Utility

Price Module through the GUI Interface to specify the

price model individually. The boundaries (min/max

values) are stored in the Business Templates which

have to be maintained by the provider for groups of

customers (e.g. well known customers, ﬁrst-use cus-

tomers, etc.) The Accounting module, which collects

from the Cloud API (e.g. no of running instances)

Open Nebula Database (e.g. history data), and from

the Cloud Resources itself (e.g. no. of SaaS users).

Figure 1: Architecture of the Cloud Utility Model.

7 CONCLUSIONS

Negotiation in form of give-and-take compromise is

an essential feature of the African Business; this made

static pricing unattractive to a typical African. Cus-

tomizing Cloud offerings for African involves embel-

lishment of the pricing scheme with negotiation that

allows the buyer to personalize pricing. This study

proposed price models that accommodates variable

pricing, hence attractive to the African markets. It

gives the customers the ﬂexibility of choosing a us-

age style accompanied by a ﬂexible pricing as well.

The customer can get better prices if the usage of the

Cloud services are predicted and therefore allows the

Cloud provider to improve its planning and therefore

improve the utilization of the Cloud infrastructure.

Effort is ongoing to develop mathematical models for

the price models and afterwards employs appropriate

simulations to validate the models.

REFERENCES

Chang, V., Wills, G., and Roure, D. D. (2010). A re-

view of cloud business models and sustainability. In

IEEE Cloud 2010, the third International Conference

on Cloud Computing, pages 43–50. IEEE Computer

Society Washington, DC, USA

2010. Event Dates:

5-10 July, 2010.

Ibrahim, S., He, B., and Jin, H. (2011). Towards pay-

as-you-consume cloud computing. In Proceedings of

the 2011 IEEE International Conference on Services

Computing, SCC ’11, pages 370–377, Washington,

DC, USA. IEEE Computer Society.

Jtmaa, J. (2010). Financial aspects of cloud computing busi-

ness models. Master’s thesis, Aalto University School

of Economics.

Ojala, A. and Tyrvainen, P. (2011). Developing Cloud Busi-

ness Models: A Case Study on Cloud Gaming. IEEE

Software, 28(4):42–47.

Osterwalder, A., Pigneur, Y., and Tucci, C. L. (1998). Clari-

fying business models: Origins, present, and future of

the concept by. Tucci Communications of the Associ-

ation for Information Systems (Volume, pages 1–25.

Rappa, M. A. (2004). The utility business model and the

future of computing services. IBM Syst. J., 43(1):32–

42.

Strømmen-Bakhtiar, A. and Razavi, A. R. (2011). Cloud

Computing Business Models Cloud Computing for

Enterprise Architectures. Computer Communications

and Networks, chapter 3, pages 43–60. Springer Lon-

don, London.

Sulistio, A., Reich, C., and Doelitzscher, F. (2009). Cloud

infrastructure & applications — cloudia. In Proceed-

ings of the 1st International Conference on Cloud

Computing, CloudCom ’09, pages 583–588, Berlin,

Heidelberg. Springer-Verlag.

Teng, F. and Magoules, F. (2010). Resource pricing and

equilibrium allocation policy in cloud computing. In

Proceedings of the 2010 10th IEEE International

Conference on Computer and Information Technol-

ogy, CIT ’10, pages 195–202, Washington, DC, USA.

IEEE Computer Society.

Zhang, Q., Cheng, L., and Boutaba, R. (2010). Cloud com-

puting: state-of-the-art and research challenges. Jour-

nal of Internet Services and Applications, 1(1):7–18.

CLOSER2013-3rdInternationalConferenceonCloudComputingandServicesScience

320