Information Technology Reference
In-Depth Information
foundations of autonomic computing to the implementation of electronic mar-
kets, and (2) simulating electronic markets for Grid and Cloud environments.
3.1 Autonomic Markets
To enable the flexibility promised by the Grid and Cloud paradigms, market plat-
forms have to be adaptive and sustainable. We argue that this can be achieved
with autonomic (self-* [13]) capabilities. Several early works served as ground-
work for prototypical implementation of autonomic aspects in complex systems.
For example, [16] discuss the need for autonomic capabilities of distributed
service systems and briefly outline the application of the self-* capabilities in
this context. Today, autonomic computing is used primarily to address techni-
cal issues to make systems autonomic, e.g. negotiation protocols to make Grid
or Cloud services self-adaptive [5] or consider autonomic service management
frameworks without explicitly considering economic methods (e.g. [12,15]).
The idea of applying economic methods and considerations to autonomic sys-
tems was initially proposed by [11]. However, current research focuses on spe-
cific economic issues and only partially considers the aspects needed to make
marketplaces autonomic. For example, [19] proposed a self-organising resource
allocation mechanism in dynamic Application Layer Networks (ALNs). They do
not, however, consider issues such as the adaptation of the market itself, but
rather the optimisation of a small piece: the allocation mechanism. Similarly,
[17] propose mechanisms that can adaptively adjust parameters based on past
participant behaviour. An example of economically-inspired market infrastruc-
tures is provided by [9] who present a self-optimising infrastructure platform
for service delivery using economic (congestion-based) pricing. Yet they, con-
sider only the infrastructure, and not the market itself. [4] study the mapping
of high-level business objectives to lower level objectives to enable autonomic
access optimisation for databases via an economic scheduler.
3.2 Simulating Electronic Marketplaces
Simulation of electronic markets for Grid and Cloud computing has been dis-
cussed in several large research projects, including SORMA [19], GridEcon [18]
and 4CaaSt [14]. [19] developed a market simulator to compare centralised and
decentralised service allocation mechanisms in market scenarios according to a
defined set of metrics. In their work, they considered complex interdependencies
that are broken down into two interrelated markets, namely a service market,
which involves trading of application services, and a resource market, which in-
volves trading of computational and infrastructure resources such as processors,
memory, etc. [18] present the GridEcon platform - a testbed for designing and
evaluating economics-aware services in a commercial Cloud computing setting.
The authors assume the diculties in predicting the context of a service market
and motivate development of an environment for evaluating its behaviour in an
emulated market platform. The platform is composed of the Marketplace, which
