The internet has allowed today's systems' degree of distribution to grow
rapidly, and thus e-learning environments have evolved from centralised,
single-point-of-entry syste,s to explore ways of realising open and global
configurations.  In this sense, the placement of Grid services is not a trivial
task.  Different related projects have assumed or proposed different service
placement configurations leading to different architectural models.  In this
deliverable we look at a learning environment through Grid lenses in order to
investigate whether today's Grid technology can be utilised to construct and
maintain such a system.

The high-level architecture of a Self e-Learning Network (SeLeNe) is first
specified at an abstract level, taking into account the set of SeLeNe services
identified in Deliverable 3.  We show that this architecture and set of
services can deliver all of the user requirements identified in Deliverable
2.2 by identifying a set of Use Cases that encompass the entire SeLeNe
functionality, and specifying a UML sequence diagram that shows how the
services interact to enact each Use Case.  A number of concrete deployment
scenarios of this architecture are then identified and evaluated. These are
(a) Centralised (b) Mediated and (c) Autonomic.  With respect to these service
placement alternatives, our aims are:

- To present how the proposed architecture relates to the efforts towards the
semantic grid. Through the development of the service based architecture we
bring forth issues that arise in relevance to today's grid technology when
examined through our metadata oriented, self e-learning network.

- To explore a number of different service placement scenarios to produce
alternative architectural models, each fitting to different learning
environments. By "learning environment", with respect to the architectural
point of view, we refer to the topology created by the actual service
providers and service consumers, the degree dynamicity, distribution and
metadata storages (i.e. where the RDF descriptions are stored, registered and
accessed). We also compare the alternatives presented with the aim of
assisting the task of constructing a SeLeNe to manage the LO descriptions'
space.