Evaluation Criteria to Run Scientific Applications in the Cloud - Cloud Computing with e-Science Applications - page 12

1.7 Cost Efficiency Evaluation

When the user decides to use a public cloud provider, it is necessary to
calculate the cost efficiency [5] of this service and if it is better to use it or buy
a cluster. To determine this, two calculations can be used, the cost efficiency
factor and the break-even point [6].

1.7.1 Cost Efficiency Factor

To calculate the cost efficiency factor for different systems, two values are
required. The first one is the cost of the cloud systems. This cost, in the
great majority of cloud providers, is expressed as cost per hour. The sec-
ond value is the overhead factor. To determine this factor, it is necessary
to execute the same workload in all the candidate systems and in the base
system.
The overhead factor O F is the execution time in the candidate system ET CS
divided by the execution time in the base system ET BS . The following equa-
tion represents this calculation:

ET
ET

CS

O =

F

Bs

As an example, we want to compare a traditional server against a machine
in the cloud. We define that the traditional server is the base system. We
need to execute the same problem on both systems and then calculate the
overhead factor. Assuming that the server takes 30 minutes to calculate and
the cloud takes 60 minutes, applying the overhead factor equation, the result
is 2 for the cloud. As the traditional system is the base system, its overhead
factor is 1.
Using the overhead factor, it is possible to determine the cost efficiency
factor CE F . The cost efficiency factor is defined as the product between the
cost per hour C HC and the calculated overhead factor, resulting in the follow-
ing equation:

CE =× CO

F C F

For example, using the calculated overhead factor 2 and assuming a cost
per hour of $5.00 of a cloud machine, the resulting cost efficiency is $10.00
per hour. The cost efficiency gives the price to perform the same amount of
work in the target system that the base system performs in 1 hour because
the cost used in our equation is the cost per hour. If the result is less than
the cost per hour of the base system, the candidate system presents a higher
cost-benefit ratio than the base system. The cost efficiency factor also can