Information Technology Reference
In-Depth Information
Consequently, there is a need for a programming model which facilitates the
specification and construction of cloud-based infrastructures. Such a program-
ming model should enable application developers to build cloud-based infras-
tructures easily and rapidly. In order to develop such a programming model,
it is desirable to have a means for mapping application requirements onto in-
frastructure products (offered by multiple providers). Many different frameworks
have been proposed for discovering and utilising cloud resources. Typically cloud
requirements are analysed from the provider's point of view, based on the re-
source capabilities offered by the provider. However, there is a mismatch between
the perspective of application developers and the low-level details supplied in a
typical resource specification. Currently, there are no suitable mechanisms for
describing requirements from an application's viewpoint. This paper proposes
an application-centric, multi-layer ontology as a way of describing application
requirements in the cloud context. The ontology allows application developers to
formulate high-level domain-specific application requirements and subsequently
to use these descriptions to search for the most appropriate set of resources in a
multi-provider cloud environment. There is potential for application developers
to utilise the proposed ontology so as to automate the resource discovery process.
This paper is organised as follow. Section 2 provides a brief summary of a
provider-agnostic programming model. Section 3 discusses related research and
defines a multi-layer model. Section 4 describes ontology translation using two
examples while conclusion are drawn in Section 5.
2 A Provider-Agnostic Programming Model
An overview of a programming model for selecting and utilising cloud-based
infrastructures in a multi-provider cloud environment is given below.
The model is wrapped in a provider-agnostic API [7] (see Figure 1) and incor-
porates a set of cloud providers that make up the cloud market. The selection
of a particular provider depends on user preferences and provider's financial
models. This set of providers is associated with a pool of available resources. In
practice, an application is mapped onto a set of resources from the resource pool
that meets the application's requirements. Multiple types of suitable resources
may be discovered. An initial result set can be filtered using heuristics in order
to find the most suitable set of resources for a particular application. After a
best fit resource is identified, it can be reserved for future use. When resources
are needed, they are instantiated and user's application is deployed. After the
application finishes, the underlying resources can be discarded.
Fig. 1.
A provider-agnostic programming model
