The main goal of the project is to provide the ability to share
information (data and meta-data) and to provide services for managing
this information in a Peer-to-Peer (P2P) environment. Data can have
different formats (documents, files, databases, etc.) and different
levels of granularity. Services are usual data management services
(storage, manipulation, querying, transformation, …).File sharing is
the « killer app » of traditional P2P systems. With RESPIRE, P2P
systems evolve to a more DBMS-oriented view, where shared resources are
more complex/structured than simple files.
The main challenges to reach the main goal of the project are the following:
1. Complex data retrieval:
the P2P approach (peer volatility, lack of a global schema) implies to
define new mechanisms for data retrieval without reusing existing
methods, usually based on a global schema and/or a centralized catalog.
2. Collaborative work support:
collaboration between peers is facilitated by grouping peers having
interest in common (for a variable duration). The main issue is to
allow creating and managing groups in a distributed way, taking into
account the dynamicity of the network and of grouping criteria. Once
groups are created, then a cooperation mechanism must be introduced so
that users of a same group can operate on data concurrently.
3. Availability of volatile resources:
as peers can enter/leave the network at any moment, replication must be
handled in a distributed way, according to network evolution.
4. Easy deployment: as peers may join the network at any moment, a service that automates the deployment of the infrastructure is necessary.
5. Dependencies handling:
the above-mentioned challenges cannot be studied separately. For
instance, replica creation is influenced by the way nodes are grouped.
Understanding those dependencies and integrating them in a global
solution is clearly a challenge by itself.
|
