Principles, Methodology and Tools for Engineering Cloud Computing Systems - Advanced Design Approaches to Emerging Software Systems - page 263

ing based on application level state, and service
level objectives considerations.

It is the latter approach that we describe here.
By providing a syntax and framework for the
definition and support of elasticity rules, we can
ensure the dynamic management of a wide range
of services with little to no modification for ex-
ecution on a cloud. With regards to the syntax,
we can identify the two following subsets of the
language that would be required to describe such
elasticity: service providers must first be able to
describe the application state as a collection of
Key Performance Indicators (KPIs), and the means
via which they are obtained in the manifest. These
will then serve as a basis for the formulation of
the rules themselves.
Alongside the syntactic requirements, a suit-
able monitoring framework must exist. A service
provider is expected to expose parameters of
interest through local Monitoring Agents, re-
sponsible for gathering suitable application level
measurements and communicating these to the
service management infrastructure via suitable

Elasticity Specification

The automated scaling of service capacity to sup-
port potential variations in load and demand can
be implemented in numerous ways. Application
providers may implement such scaling at the ap-
plication level, relying on an exposed interface of
the cloud computing infrastructure to issue spe-
cific reconfiguration requests when appropriate.
Alternatively, they may have a desire to keep the
application design free of infrastructure specific
operations and opt instead to delegate such con-
cerns to the infrastructure itself. With a sufficient
level of transparency at the application level for
workload conditions to be identified, and through
the specification of clear rules associating these
conditions with specific actions to undertake, the
cloud can handle dynamic capacity adjustment on
behalf of the service provider.

Figure 5. Overview of an OVF processing engine plus additional machinery for service provision