Information Technology Reference
In-Depth Information
allows trading goods using different market mechanisms, and the Workflow En-
gine, which enables a simple composition of a market environment by describing
the service interactions between economics-aware services. [14] discuss a mecha-
nism for the resolution of the customers' requirements that enhances the process
of selecting Cloud services from the business point of view. The work is related
to the 4CaaSt project and aims to create a PaaS Cloud platform that supports
the optimised and elastic hosting of Internet-scale multi-tier applications.
[2] discuss a framework for modelling and simulating service-oriented applica-
tions and autonomic policies for service provisioning and resource orchestration
for Application Layer Networks in utility computing environments. The approach
is evaluated within CATNETS project and investigates the use of an economic
model (Catallaxy) in distributed environments like Grids and P2P networks. [20]
discuss the design of a simulator with a set of features for simulation of Grid
testbeds as an extension to GridSim. They model heterogeneous computational
resources of variable performance, scheduling of jobs based on various policies,
differentiated network service, and workload trace-based simulation.
Although many of these market simulators successfully address some of the
main challenges of electronic markets in distributed environments, they are fairly
static and do not have any autonomic capabilities. Therefore to orchestrate and
evaluate autonomic markets, a more flexible simulation approach is necessary.
4 A Case Study for Market Simulation Using GridSim
In [6], we used GridSim as a means to explore how a market could be monitored
as first steps towards adding autonomic capabilities to an electronic market. We
selected GridSim for a variety of reasons: (1) it implements numerous mecha-
nisms for resource allocations [3] as well as interfaces for implementing addi-
tional mechanisms; (2) it is designed as an extensible multi-layer architecture
which allows new (technical) components or layers to be added and integrated
[7]; (3) it allows different classes of heterogeneous resources; and (4) as an open-
source toolkit it has already been used widely. Although GridSim simulates Grid
resource and networks and not the Cloud computing paradigm directly, it is im-
portant to note that these two contexts do not differ significantly, as the core
techniques for matching buyers to sellers are equal and independent of techni-
cal paradigm. Using the layered architecture of GridSim, we implemented three
monitoring sensors for: market mechanisms, the market in general, and com-
putational infrastructure of the platform as extensions to the existing GridSim
layers to monitor the infrastructure and institutional performance of the market.
The
Mechanism sensor
monitors the performance of a market mechanism,
which includes revenue, the number of resource allocations, and average price
for a single unit of resource. The actual allocation is handled by GridSim. The
mechanism sensor uses the GridSim interfaces to receive a notification of an allo-
cation,i.e.amatchofabidtoanask.Once a resource is allocated, the sensor re-
ceives and stores information about the allocation in the knowledge component.
Using the same GridSim interfaces, the sensor also gathers mechanism-specific
information like number of bids and asks awaiting allocation.
