Information Technology Reference
In-Depth Information
It is important to note that the total execution time could be reduced in
this case by launching one machine per file, but that decision would imply a
considerable cost increase.
6.5.3 e-Clouds IaaS Provider
One of the main goals for e-Clouds is to support multiple underlying infra-
structure providers. This means that the user would be able to choose
between different alternatives and select the one that best fits personal
needs. Because there are no well-defined standards among cloud suppliers,
this platform independence requires some extra development work from
the e-Clouds team and is why this first version release supports just one of
them. Taking all this into account and with the objective of launching a first
version as soon as possible, the e-Clouds team selected AWS and Heroku for
the initial testing.
Amazon Web Services has consolidated as one of the biggest and most
complete public clouds offering IaaS. Its low prices and the flexible resource
configuration are ideal for an initial testing of the e-Clouds ideas. Also,
because of its large number of users and success cases, there is complete
documentation available regarding performance benchmarking and testing
under high-performance scenarios [34, 35].
Heroku is a cloud platform as a service (PaaS) offer for fast and simple
deployment of web applications [36]. It actually supports Ruby, Node.js,
Clojure, Java, Python, and Scala applications. As part of its proposal, Heroku
has a free usage tier specially designed for developers. This free tier allows
users to run testing environments in a basic configuration of one web server
and an SQL database. Deployment can be easily done through a Git reposi-
tory configuration and a set of command line tools.
As shown in Figure 6.8, e-Clouds web portal and RM are initially deployed
in Heroku for ease. This includes the presentation layer developed under
Ruby on Rails and the relational database running in PostgreSQL. On the
other hand, AWS is used to run scientific workloads, store user files, and com-
municate information through reliable queues. A more detailed description of
each one of these components can be found further in this chapter.
6.5.4 Monetization
As in any cloud solution, there is a cost transference between the infrastruc-
ture usage and the service delivered. During this initial phase, e-Clouds
only charges each individual user by the different resources the user effec-
tively consumed, including storage, computing hours, and communication.
Although the total cost of an execution depends on these three basic factors,
the charging model for each of them is slightly different.
Search WWH ::
Custom Search