Dynamic coordination rules in peer-to-peer database

In existing peer-to-peer database framework designs, coordination rules are assumed already present and never changed during the whole course of operation. This paper investigates how coordination rules are created and changed, hence helping ease the procedure. Local database can be on and off dynamically, but this feature of P2P database is inconsistent with fixed coordination rules, for dependency path will be broken when an intermediate peer is absent. A restoration mechanism is designed in this scenario to realize dynamic coordination rule. To achieve this, coordination rules on the same dependency path have to be available after the path is broken, and combined together to form a new dependency path and bypass the absent peer. To backup rules before host is down they can be published as resource advertisement to remote peers by underlying P2P platform facility. Actually since coordination rules are no longer bounded with their host, they can be viewed independent from the database system to form a coordination rule P2P network, with some peers having no database and purely as rule cache. The protocols about rule cache, combination and new rule creation request in such network are discussed. Rules float along dependency paths across network and combine to form a new rule where necessary. A peer wanting to create new coordination rules can publish query and if there is a rule on another peer which can be combined with the existing one, a new rule is created and send back. This dependency path discovery process can be similar to route discovery process.