Information Technology Reference
In-Depth Information
by monitoring sensors, which gather low-level monitoring data from the market
middleware and implementation of the market model. Using the predefined map-
pings, the measured monitoring data is mapped to the higher level performance
indicators, which is then used to assess the market performance.
The main purpose of a
knowledge
component is to store, manage and anal-
yse real-time monitoring data and experiences from previous adaptations. The
knowledge gathered in this process can be (1) empirical, i.e. derived from the
observations on the market (e.g. infrastructure status, payoffs from previous
adaptations, etc.), (2) contextual, i.e. instance-specific (e.g. initial/desired con-
figurations and business models), and (3) institutional, i.e. concerning the eco-
nomic anatomy of the marketplace (e.g. valid alternative configurations and
market rules, constraints and regulations). While empirical knowledge is gath-
ered through monitoring and logging techniques, contextual knowledge is (ini-
tially) set by the market administrator. Institutional knowledge is defined
partially by the market administrator and partially established based upon con-
textual knowledge and changes (i.e. evolutions) within the marketplace.
The
analysis
phase is used to analyse mapped data from the monitoring sen-
sors to derive possible actions for market adaptation in order to improve market
performance with respect to a set of goals. As already mentioned, there are two
main adaption options: the market's infrastructure and its configuration. Find-
ing which of these options is the most fitting is, however, not trivial. Autonomic
adaptation of infrastructure properties has already been discussed in a large
body of literature (e.g. [1,8,10]). This, however, is not the case for institutional
adaption. To facilitate institutional adaption, we need to understand what dif-
ferent market configurations mean for the fulfilment of a given set of goals, which
can be achieved through simulation to enable the analysis of what-if scenarios
to determine and assess adaption options.
The
planning
phase of the autonomic adaptation cycle includes two impor-
tant steps. Firstly, it identifies the most suitable adaptation path(s) for the exe-
cution of the infrastructure and/or institutional changes by leveraging contextual
knowledge. Secondly, as an adaptation path may include more than one market
component or steps, it determines the order and timing of the adaptations to be
instrumented. This may result in multiple rounds of the adaptation cycle with
the goal of observing how single changes impact the market performance and
ultimately lead to an iterative adaption process.
The
execution
phase is the execution of an adaption path. In the case of
an infrastructure adaption, this relates to an interaction with the resource fab-
rics through the platform middleware. For institutional adaption, it refers to a
check point of the current market status, a new parameterisation of the market
configuration, and redeployment (if necessary) of effected market components.
3 Related Work
For positioning our work within the state-of-the-art, we briefly describe existing
research on electronic markets and classify it into two categories: (1) applying
