Scalability
and Fault-Tolerance in an Event Rule Framework for Distributed Systems
Abstract
In Event/Rule Framework (ERF) for Distributed Systems (DS), events, rules and
services are treated as objects, and they are used as abstractions for
specifying system behavior [Arroyo02b] Rules are triggered when events match the
rules' event pattern, [Arroyo00a] and they can execute actions, which can be
either event postings or remote method invocations.
ERF lacks of scalability and fault-tolerance: (a) performance is decreased as
the number of rules increases; (b) since the execution of the DS depends on
RUBIES, and (c) since rule actions may invoke methods from remote services, then
unavailability will cause the whole DS to be down; and (d) such dependency on a
single service may pose a scalability problem. For achieving scalability and
fault tolerance in ERF, one approach is distribution and replication.
Scalability can be achieved by distributing rules over several replicated RUBIES
instances; fault tolerance can be achieved by means of RUBIES replication and
replicating the corresponding rule sets. The architecture of this approach is
presented in this paper.
Full
document is downloadable from:
| 
