Next: 3.1 Problem Statement
Up: FastReplica: Efficient Large File
Previous: 2 Additional Related Work
3 FastReplica Algorithm
In this Section, we describe the formal problem definition and
introduce the new algorithm FastReplica for replicating the large
files in the Internet environment.
In Section 3.2, we introduce a core (an induction step) of the
algorithm, called FastReplica in the small, and demonstrate its
work in the small-scale environment, where a set of nodes to which a
file has to be replicated is rather small and limited, e.g. 10 - 30
nodes.
In Section 3.3, we perform a preliminary analysis of
FastReplica in the small and its potential performance benefits, and
outline the configurations and network conditions when
FastReplica may be inefficient.
In Section 3.4, we describe the general, scalable algorithm,
called FastReplica in the large, and demonstrate its work in the
large-scale environment, where a set of nodes to which a file has to
be replicated can be in a range of hundreds/thousands of nodes.
In Section 3.5, we extend the FastReplica algorithm
to be able to deal with node failures.
Subsections