Hedging Cloud Energy Costs via Risk-free Provision Point Contracts
Owen Rogers
and Dave Cliff
Department of Computer Science, University of Bristol, Bristol, U.K.
Keywords: Pricing, Derivatives, Energy Market, Electricity Futures.
Abstract: The cost of electricity is a major concern to public providers of cloud computing services. On-demand
pricing, common amongst cloud providers, does not aid the provider in planning future demand and
therefore purchasing energy at discounted rates. In this paper, we describe a number of advance pricing
schemes for cloud computing resources based on provision-point contracts, commonly used by deal-of-the-
day websites such as Groupon. We propose three models – Group Provision Points, Contributory Provision
Points, and Variable Reward Forwards – that each reward consumers with reduced prices for advance
reservations, while allowing providers to make accurate forecasts of energy usage. Furthermore, we show
how the schemes are risk-free for the provider, guaranteeing to be at least as profitable as on-demand
schemes. We present results from a simulation of the schemes, and compare the results to our analytically
derived predictions.
1 INTRODUCTION
Consumers of cloud computing resources typically
pay a single price to access a virtual machine for a
specified period of time. This single price covers the
virtual machine’s fraction of the cost of the physical
server itself, maintenance and repairs, the physical
datacentre space, the electricity needed to power it,
and the cost of air conditioning to cool the
datacentre.
In on-demand pricing, consumers gain access to
the resource immediately and are charged for the
amount of time they use the resource.
In forward pricing, consumers gain access to the
resource at a specified time in the future, and have
access for a pre-agreed duration.
Air conditioning and datacentre space are
generally fixed costs. Regardless of how many
servers are placed in the datacentre, these costs will
essentially be the same.
Electricity costs for powering servers are
variable costs. The total electricity required by the
provider is proportional to the amount of virtual
machines demanded by the provider’s customers.
Energy costs are a significant cost for providers
of public cloud computing resources.
Estimates for the contribution of server
electricity to the total cost of ownership (TCO) of a
physical server vary between 3% and 15% (Barroso
and Hölzle 2009; Berl et al., 2009). Volume servers
account for 34% of datacentre electricity usage
(Brown, 2008). A full review of datacentre costs can
be found in (Patel and Shah, 2005).
This cost therefore impacts the price paid by
consumers to access virtual machines, and the profit
achieved by the provider. In a competitive
marketplace, keeping prices as low as possible is
critical for commercial success.
Currently, research is being focussed on reducing
the power consumption of computing technology
(Barroso and Holzle 2007; Lee and Zomaya, 2010).
The primary focus of this research is reducing
carbon footprint, but reducing expenditure is an
important factor too.
Typically, a cloud provider would purchase
electricity on-demand for a fixed price to power its
datacentre. This could be directly from an energy
supplier, or from a broker who hedges market-traded
instruments to offer fixed prices to its clients.
Larger cloud providers might purchase electricity
directly from the spot-market, where prices vary
over time to match supply with demand. These
larger providers may also generate their own
electricity and be able to contribute energy to the
grid as well as consuming it through bilateral
agreements.
Some research has been directed at moving
virtual machines between datacentres with the aim
244
Rogers O. and Cliff D..
Hedging Cloud Energy Costs via Risk-free Provision Point Contracts.
DOI: 10.5220/0004374502440252
In Proceedings of the 3rd International Conference on Cloud Computing and Services Science (CLOSER-2013), pages 244-252
ISBN: 978-989-8565-52-5
Copyright
c
2013 SCITEPRESS (Science and Technology Publications, Lda.)
of finding the cheapest spot-price.
Qureshi et al. were the first to propose
dynamically assigning computational workloads in
distributed systems to locations where electricity
may be cheaper. They found savings of millions of
dollars could be achieved through a simulation
(Qureshi et al., 2009).
A similar method was suggested by Rao et al.,
but the dynamic allocation also takes into account
the latency between different locations, so that QoS
metrics would be met while electricity cost reduced
(Rao et al., 2010).
Buchbinder et al. extended these methods so that
only batch applications would be migrated to
cheaper markets (Buchbinder et al., 2011). In this
way, applications that could tolerate a delay would
use the cheapest electricity, and interactive
applications would not cause poor user-experience
as a result of the overhead involved in migrating the
application
Ding et al. also proposed that virtual machines
could be moved between datacentres based on
electricity prices (Guo et al., 2011).
However, little research has been conducted on if
providers can use derivative contracts to purchase
electricity in advance for a discount.
The cloud provider could potentially decrease its
costs by purchasing electricity futures directly (Hull,
2008). Futures contracts are a type of derivative that
give buyers guaranteed access to the resource in
advance of when it is delivered: the user is obliged
to take ownership of the resource on the delivery
date that the contract specified. The provider could
then engage a broker to provide fixed-price
electricity to top up its pre-bought electricity
capacity
A futures contract typically details the size of the
commodity being purchased. In electricity futures,
the commodity is a quantity of electricity delivered
for a fixed period of time, typically a month or a
quarter.
However, the use of electricity futures can have
significant associated risks. If the provider invests in
a future which is subsequently not fully utilised by
customers, then it is possible it will not cover the
investment. Electricity delivered to the cloud
provider cannot be stored; if it is not used as it is
delivered, then it is wasted.
Considering an electricity future for one months
delivery of 1MW costs over $35,000, this risk can be
sizeable
1
.
1
ICE UK Base Electricity Futures, November 2012
In this paper, we propose three pricing schemes that
allow the provider to purchase electricity futures
with no-risk that they will subsequently fail to utilise
their investment effectively. The provider is
guaranteed to be at least as profitable as using a
traditional on-demand pricing scheme.
Our schemes are based on provision-point
contracts (also known as assurance contracts). In a
provision-point mechanism, members of a group
pledge to contribute to an action if a threshold of
some order is met. If this threshold is met, the action
is taken and the public goods are provided;
otherwise no party is bound to carry out the action
and money paid is refunded (Bagnolli and Lipman
1989).
Such a mechanism is used by deal-of-the-day
website Groupon
2
. Users make requests for special
offers by purchasing a coupon. When a threshold is
reached, the deal is profitable to the provider and the
offer is confirmed.
In previous work, we showed how provision-
point contracts can be used to schedule virtual
machines more effectively on a large-scale cloud
infrastructure (Rogers and Cliff, 2012; Rogers and
Cliff, 2012).
In this paper, we amend traditional provision
points by changing the beneficiaries of the contract
and the value of the offer to create a number of new
pricing schemes.
Consumers of cloud computing resources can
purchase these in advance for discount, while
retaining the ability to purchase additional resources
on-demand. The cloud provider subsequently uses
this information to purchase electricity futures.
We show how Group Provision Points,
Contributory Provision Points, and Variable Reward
Forwards allow providers to make accurate forecasts
of energy usage and therefore reduce their costs
through the purchase of electricity.
We present results from a simulation of the
schemes, and show that our schemes have benefits
for both provider and consumer compared to
traditional on-demand and forward pricing.
2 PRICING SCHEMES
2.1 On-demand Pricing
In standard on-demand pricing there is a period of
duration N intervals, where resources are purchased
and then immediately available.
2
www.groupon.com
HedgingCloudEnergyCostsviaRisk-freeProvisionPointContracts
245
The provider charges customers a cost C
o
to use the
computing resource for an interval i. The total
demand experienced for resources in time interval i
is t
i
. In this case, the total revenue (REV) achieved
by the provider over the period is the total demand
experienced at the on-demand price:
The provider will be required to pay for
electricity for the duration of the interval that the
virtual machine is running at a cost E
o
from the
energy supplier or broker. The electricity required
per virtual machine for the interval is β. In this case,
the cost of electricity (COE) to the provider is the
total demand experienced, at the cost of on-demand
electricity per virtual machine:
Therefore, the provider’s profit using an on-
demand model is:
2.2 Forward Contracts
Consider a pricing model for cloud computing which
uses two periods, each period consisting of N time
intervals.
In the first period, ‘the reservation period’,
consumers purchase advance reservations (or
forwards) at a cost C
r
, which allows them to use a
resource at a specific interval i during the next
period. The total number of resources reserved in a
time interval i is r
i
.
In the second period, “the execution period”,
consumers gain access to their reservations at the
specified time interval. Consumers may also
purchase access to a resource for the duration of an
interval at a cost C
o.
The total demand experienced
for resources in time interval i is t
i
In this case, the revenue achieved over the period
is the sum of reserved resources bought at the
reserve price, plus the additional resources bought
on-demand at the on-demand cost:
As the provider has committed to deliver a number
of resources through the sale of forward contracts on
computing resources, she can use this information to
purchase forward contracts on electricity to obtain a
saving on consumption. The provider can choose to
buy θ forward electricity contracts, where each
contract entitles them to use I units of electricity for
a period of N time intervals at a cost E
r
per time
interval.
The cost over the period is the cost of purchasing
reserved electricity across the entire period, plus the
sum of the cost of purchasing on-demand electricity
required in addition to the reserved electricity.
Therefore the profit obtained via hedging
electricity consumption through the use of forward
contracts on electricity is:
,
For the model to be worth implementing for the
provider, it must offer a greater profit than using an
on-demand model:
∑
However, for the model to be beneficial to the
user, the user must be incentivised to provide a
forecast. Therefore, the cost of reserving a resource
must be less than the cost of buying a resource on-
demand:
So our conditions for the model to be beneficial to
all parties are:
1
∑
2
With forward pricing on computing resources, the
provider might choose to fix C
o
and C
r
so that
customers are fully aware of the pricing they will be
charged. In this case condition (1) is satisfied, and
consumers will use the service.
CLOSER2013-3rdInternationalConferenceonCloudComputingandServicesScience
246
However, as condition (2) is dependent on
∑
,
the provider is not aware of if the scheme will be
more profitable than on-demand pricing until all
users have purchased forward contracts and the
provider must deliver the resource.
The provider must provide users with access to
their reserved instances for smaller cost (and
therefore less revenue), but may not benefit from
cheaper electricity costs in all cases.
2.3 Group Provision Points (GPP)
This issue can be circumvented with a provision
point contract. We now introduce an additional,
intermediate phase – the ‘confirmation phase’:
1. Reservation Phase: Users request resources to
be consumed in the execution phase
2. Confirmation Phase: If the provider finds that
they will benefit as a result of the model by
conditions (1) and (2) being met, they will
confirm user’s requests and the contracts are
confirmed. If either condition is not met, all
contracts are cancelled.
3. Execution Phase: Users gain access to their
confirmed resources, and may also buy
additional on-demand resources.
If the requirements of the user population are
found not to produce an increase in profit, the
provider cancels all contracts and no revenue is lost
as a result. If the scheme is profitable, all contracts
are confirmed. This is equivalent to a traditional
provision-point contract used by deal-of-the-day
websites such as Groupon.
2.4 Contributor Provision Points (CPP)
The forward and GPP schemes are extremes. In the
forward scheme, all users who submit a reservation
benefit from reduced prices, in spite of it sometimes
not benefitting the provider. In the GPP scheme
either all, or no, users benefit from reduced prices
depending on whether an advantage is gained by the
provider or not.
A compromise might be to only confirm contract
requests to the consumers that contribute to the
purchase of advanced electricity during the
confirmation phase. This could be based on the
earliest consumers who request a reservation.
Customers who submitted a late reservation would
have their contract cancelled, as their discount
would not contribute to cheaper electricity.
The provider would typically determine how
much advance electricity θ to purchase based on
some function of the profile of the reserved
resources over the month.
⋯
If the provider chooses to purchase forward
contracts on electricity, this will provide the
provider with units of electricity each interval for
N intervals. Therefore, the total electricity available
to the provider over the period is. This will
support q contracts:
1
If we confirm only q contracts, and cancel all
others:
Substituting into (2):
1
3
The vulnerability of the forwards has now been
removed, as the conditions for profitability no longer
depend on the uncontrollable number of
reservations.
As long as prior to implementing the scheme
conditions (1) and (3) are met and E
r
is set to be the
maximum likely cost of an electricity future, the
scheme will generate a profit over on-demand
pricing.
This scheme also protects the provider against
changes in the cost of electricity forwards. If the cost
of a forward does not satisfy the following, the
provider should cancel all contracts:
2.5 Variable Reward Forwards (VR)
In the variable reward model, consumers are given
the guarantee that when purchasing a forward in the
reservation period, the price payable for the forward
will be the same, or less, than the cost of an on-
demand resource. The exact value of C
r
is not known
until the execution period and is determined on a
profit-sharing basis, where μ is the share desired by
the provider.
HedgingCloudEnergyCostsviaRisk-freeProvisionPointContracts
247
∑
Users pay the minimum C
r
to be profitable, plus
a share of the saving achieved:
1
If C
r
> C
o
then on-demand instances are cheaper
than reserved and the model will fail. In this case, no
discount is to be offered and C
r
= C
o
. The condition
for this is:
0
,0
1, 0
This will always be as least as profitable as on-
demand instances as users pay the on-demand price
if no saving can be made.
3 SIMULATION
3.1 Setup
A simulation was written in Python, the primary
aims being to verify that the models outperform
conventional on-demand and forward pricing
schemes when applied to practical applications, and
that users can make savings using a rational
approach to forecasting. Furthermore, a simulation
will aid comparing models where the cost of
electricity futures varies over time.
In our first simulation, we wish to determine which
contract model generates most profit in a monopoly
market where the broker is the only (or at least the
preferred) provider of cloud resources. Our objective
is to understand the profitability implications for the
provider of such schemes, and the cost implications
for the consumer.
For electricity, we assume that the broker may
purchase electricity futures for a period of a calendar
month, which supplies 1MWh of electricity per
hour. We obtain prices of ICE UK Base Electricity
Futures over a 39 month period from March 2012
(Figure 1). The cost of electricity on-demand from
the grid is £0.01/kWh, based on (Barroso and Hölzle
2009) which is the most reliable source of this
information in current academic research.
Figure 1: Cost over electricity futures over time.
Group provision points (GPP), contributory
provision points (CPP) and standard forwards all
have the same prices for reservations and on-demand
instances. An on-demand instance is set at
£0.01/Computational-Unit/hour, which is a
reasonable figure in the current market.
The contributory provision point provider
believes that the maximum she will need to pay for
electricity in the foreseeable future is £55,
Therefore,
0.007863
The CPP provider sets C
r
= £0.007864. This will
guarantee her a benefit over on-demand pricing as
long as electricity does not go higher than £55.
The forward and GPP providers set C
r
to be the
same. The variable reward forward (VR) provider
sets C
o
= 0.01, but has no basis for determining a
reserved price. She decides that she requires 50% of
any saving used by the scheme to be retained as
profit, and the other 50% to be split to consumers
who reserved.
We simulated a demand curve varying over time
using a combination of 5 types of users:
Flat profile represents where demand is
constant, and hence trivially easy to predict;
Random profile represents stochastically
unpredictable demand, chosen randomly from a
normal distribution;
Sine profiles (with period of 24 hours) are an
approximation to daily rhythms, where demand
varies sinusoidally, peaking in the middle of
the day and at a minimum in the middle of the
night. More precisely, in our simulations this
sinusoidal demand pattern peaks around mid-
day, and demand can never be negative, so a
function of the form 1+cos(2πh/24) is used,
where h is the hour-number in the day. We
have explored three variations of these sinusoid
patterns:
CLOSER2013-3rdInternationalConferenceonCloudComputingandServicesScience
248
a. Flat Sine represents constant a constant
baseline of demand with periodic
variations across each day;
b. Growing Sine represents daily periodic
demand, with the baseline increasing
steadily across the month;
c. Shrinking Sine represents daily periodic
demand, shrinking through the month.
We create a demand curve by combining
different quantities of these users such that demand
is generally growing over time so that the benefit of
purchasing additional electricity futures can be seen
in our results. There are 2000 users in total using the
simulation.
The aim of the demand curve is to determine if
the scheme can be profitable in a heterogeneous
market of different users with different demands,
which follows an increasing trend. We are not aware
of any real-world data on public cloud demand
which we could use over such timescales, so this is a
suitable approximation in this preliminary study.
We assume the provider has servers that can
support 8 virtual machines, and each server uses
380W.
3.2 Results
3.2.1 Provider Cost Reduction
For clarity, figures 3-8 show data points averaged
over the last 2 months with the corresponding
standard deviation shown in error bars.
Figure 2: Profit achieved using pricing models.
Figure 2 shows that the CPP model does generate
more profit for the provider than on-demand pricing
alone. Initially this is around a 10% increase, but this
decreases as forward electricity prices increases. In
month 17 (hour 11424), electricity futures rise above
the point where our reserved pricing is profitable,
and so all contracts are cancelled. This can be seen
in Figure 3 where no energy resources are reserved
as the cost of electricity forwards goes higher than
our reserved pricing threshold.
Figure 3: Power used by CPP model.
In this case, the profit achieved is the same as
on-demand pricing. In Figure 4 it can be seen at this
point that no contracts are confirmed, and all users
must purchase on-demand resources.
Figure 4: Resource allocation in CPP model.
The VR model is the most profitable for the
provider, being up to 16% more profitable than on-
demand pricing. This is because the provider is not
committed to giving a specific discount to the
consumer. The fact that any benefit obtained through
advance reservations is shared means that the
provider gains when big savings are achieved, and
doesn’t lose out when a loss is likely. The VR model
is also not negatively impacted as a result of changes
in the price of electricity futures and, unlike the CPP
model, generates more profit than on-demand during
these price hikes. Figure 5 shows the purchase of
electricity by the provider. Figure 6 shows the
purchase of resources by consumers.
HedgingCloudEnergyCostsviaRisk-freeProvisionPointContracts
249
Figure 5: Power used by forwards and VR model.
Figure 6: Resource allocation using forwards and VR.
Forwards are generally less profitable than on-
demand resources, by quite a large margin. Clearly,
the pricing is too low for purchasing advance
reservations to be profitable, but determining this
price is not easy as the number of reservations is not
known until they have all been requested.
Figure 7: Power used by GPP model.
Figure 8: Resource allocation using GPP model.
The GPP model also fails to deliver significant
gains in profit. The model protects losses as a result
of not giving discounts when forwards are less
profitable than on-demand, but it doesn’t achieve
high profits when forwards are more profitable as
everyone receives the discount (figures 7 and 8).
3.2.2 Provider Cost Reduction
Forwards are generally the most beneficial to the
consumer achieving a mean saving of around 20%
the cost of an on-demand instance, and reducing
costs for all market demand profiles (Figure 9). This
is because the consumer always gains access to the
resource, and thus their net costs are reduced. The
mean price does not equal the cost of the reserved
resource in all situations because sometimes a user
purchases a resource that subsequently she does not
require, but which she has already agreed to pay for.
Figure 9: Mean price per computational unit using
forwards.
The CPP model rewards consumers with around
a 15-10% saving when a cost saving is achieved,
CLOSER2013-3rdInternationalConferenceonCloudComputingandServicesScience
250
with this rising to no discount when electricity prices
increase (Figure 10). All consumers make a saving
using the CPP model.
Figure 10: Mean price per computational unit using CPP.
Figure 11: Mean price per computational unit using VR.
Figure 12: Mean price per computational unit using GPP.
The VR model provides a discount of around 5%
on average (Figure 11). However, users with random
demand profiles spend more using the scheme than
if using just on-demand resources. This is as a result
of poor predictability, which results in the purchase
of advance resources which are subsequently not
used.
The GPP model is generally unattractive to
consumers (Figure 12). Only occasionally is a
discount awarded, and it is unlikely this would not
occur enough to be of interest.
4 CONCLUSIONS
In this paper, we have introduced and analysed a
number of novel pricing schemes for cloud
computing which we have shown to offer
opportunities for increasing profits by reducing the
cost of purchasing electricity.
Group Provision Points are unlikely to be
implemented in a commercial offering, as the
scheme does not take full advantage of information
acquired through the sale from consumers.
Consumers do not receive regular enough discounts
to make forecasting worthwhile, nor does the
provider benefit from reduced electricity costs.
We believe Contributory Provision Points and
Variable Reward Forwards are the most attractive of
the schemes discussed. Contributory Provision
Points will favour those who can predict their future
demands earlier. Variable Reward Forwards gives
everyone who contributed to a reduced cost with a
share of the saving. It is likely Variable Reward
Forwards would be seen as fairer by the user-base as
everyone is rewarded; not just those who contribute
to the discount, which cannot be established
beforehand.
Both of these schemes can be configured to
outperform on-demand pricing by setting reserved
pricing appropriately.
However, because of the size of the electricity
futures involved, only larger providers would be
able to take advantage of the schemes.
Further investigation should be conducted on
how these schemes can be used in bilateral
arrangements, where datacentres may produce their
own electricity which may be ploughed back into the
electricity grid. Furthermore, can these schemes be
enhanced through the use of cloud spot-markets, or
reserved instances?
In this work, we assumed air conditioning was a
fixed cost which doesn’t change with increasing
number of servers. However, a gradual increase in
air conditioning energy is likely to be seen as a
result of the increased heat generated by servers. In
HedgingCloudEnergyCostsviaRisk-freeProvisionPointContracts
251
future work, including air conditioning costs in the
model could further reduce expenditure.
A significant amount of work is still required to
determine if these schemes can be implemented
commercially. In future work, we plan to create a
simulation of a competitive market of providers
utilising the scheme. Our objective is to see if one
scheme becomes dominant in the marketplace. We
also wish to investigate if the providers can change
pricing with a view to acquire more business. This
could eventually lead to a market for provision point
contracts in cloud computing.
ACKNOWLEDGEMENTS
We thank the Large-Scale Complex IT Systems
Initiative (www.lscits.org) as well as HP Labs
Adaptive Infrastructure Lab for providing additional
financial support.
REFERENCES
Bagnolli, M, and B Lipman. 1989. “Provision of Public
Goods: Fully Implementing the Core Through Private
Contributions.” Review of Economics Studies 56: 583–
601.
Barroso, L. A., and U. Holzle. 2007. “The case for energy-
proportional computing.” Computer 40(12): 33–37.
http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4
404806 (November 30, 2011).
Barroso, Luiz André, and Urs Hölzle. 2009. “The
Datacenter as a Computer: An Introduction to the
Design of Warehouse-Scale Machines.” Synthesis
Lectures on Computer Architecture 4(1): 1–108.
http://www.morganclaypool.com/doi/abs/10.2200/S00
193ED1V01Y200905CAC006.
Berl, a., E. Gelenbe, M. Di Girolamo, G. Giuliani, H. De
Meer, M. Q. Dang, and K. Pentikousis. 2009.
“Energy-Efficient Cloud Computing.” The Computer
Journal 53(7): 1045–1051. http://
comjnl.oxfordjournals.org/cgi/doi/10.1093/comjnl/bxp
080 (November 7, 2012).
Brown, R. 2008. “Report to congress on server and data
center energy efficiency: Public law 109-431.”
http://escholarship.org/uc/item/74g2r0vg.pdf (January
21, 2013).
Buchbinder, N., N. Jain, and I Menache. 2011. “Online
job-migration for reducing the electricity bill in the
cloud.” NETWORKING 2011. http://
www.springerlink.com/
index/86060362741H6X88.pdf (January 23, 2013).
Guo, Yuanxiong, Zongrui Ding, Yuguang Fang, and
Dapeng Wu. 2011. “Cutting Down Electricity Cost in
Internet Data Centers by Using Energy Storage.” In
2011 IEEE Global Telecommunications Conference -
GLOBECOM 2011, IEEE, p. 1–5. http://
ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=61
34209 (January 23, 2013).
Hull, John C. 2008. Fundamentals of Futures and Options
Markets. 6th ed. Prentice Hall.
Lee, Young Choon, and Albert Y. Zomaya. 2010. “Energy
efficient utilization of resources in cloud computing
systems.” The Journal of Supercomputing. http://
www.springerlink.com/index/10.1007/s11227-010-
0421-3 (July 18, 2011).
Patel, C. D., and A. J. Shah. 2005. “Cost model for
planning, development and operation of a data center.”
Development 107: 1–36. http://www.hpl.hp.com/
techreports/2005/HPL-2005-107R1.pdf (November
30, 2011).
Qureshi, Asfandyar, Rick Weber, Hari Balakrishnan, John
V. Guttag, and Bruce Maggs. 2009. “Cutting the
Electric Bill for Internet-Scale Systems.”
http://dspace.mit.edu/handle/1721.1/62585 (January
21, 2013).
Rao, Lei, Xue Liu, Le Xie, and Wenyu Liu. 2010.
“Minimizing Electricity Cost: Optimization of
Distributed Internet Data Centers in a Multi-
Electricity-Market Environment.” In 2010
Proceedings IEEE INFOCOM, IEEE, p. 1–9.
http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5
461933’ escapeXml='false'/> (January 7, 2013).
Rogers, O., and D. Cliff. 2012. “Options, forwards and
provision-point contracts in improving cloud
infrastructure utilisation.” Journal of Cloud
Computing: Advances, Systems and Applications
1(21).
Rogers, Owen, and Dave Cliff. 2012. “The Use of
Provision Point Contracts for Improving Cloud
Infrastructure Utilisation.” In Grid Economics and
Business Models,.
CLOSER2013-3rdInternationalConferenceonCloudComputingandServicesScience
252
