QoS-Oriented Service Computing - Advanced Design Approaches to Emerging Software Systems - page 294

Figure 7. On-demand resource provisioning through zone migration. If 'zone0' containing 5 players is
migrated to the idle local session, the workload then falls down to 0 and the running total of the measure-
ment data of RTFThroughputIn and RTFThroughOut remain unchanged. If 'zone0' is migrated back to
original local session, the running total of measurement data of RTFThroughputIn and RTFThroughOut
start increasing again

can be complex as it involves an infrastructure
spanning multiple autonomous administrative
domains of service providers. The SLA manager
security model and associated policies can have
specific architecture depending on the granularity
of the domain system security requirements, which
goes beyond the scope of this paper.
The reference architecture is proposed based
on the requirements identified from SLA related
research projects. It aims to provide an insight into
a comprehensive SLA manager, and to present a
working model of SLA manager starting from
service provider publishing the service to billing
the service consumer at the end-of-life SLA. It
integrates most of the functional aspects during
the SLA life cycle. However, some of components
/sub-modules are not mandatory and may not
be needed subject to the specific requirements.
Informed by the research findings from the sur-
vey, the reference architecture also supports the

incorporation of domain ontology to facilitate the
semantic-aware SLA manager.
The proposed reference architecture recom-
mends using OGF standard WS-Agreement for
SLA representation. WS-Agreement has been
widely used in Grid /e-Science communities. It is
an extensible language and leaves open space for
defining domain specific terms. This means the
defined domain-specific ontology can be 'embed-
ded' into the WS-Agreement based specification.
A generic SLA manager software framework
can be developed based on the proposed reference
architecture, to provide prefabricated software
building blocks that developers can use, extend,
or customize for specific SLA life cycle manage-
ment solutions.
At the time of writing this chapter, the work-
shop “ The future of Cloud Computing ” held in
January, 2010, (http://cordis.europa.eu/fp7/ict/
ssai/events-20100126-cloud-computing_en.html)