RESUMEN
Distributed configurations, where nodes may act
independently, offer potential for fault tolerance through data replication. On
the other hand, they imply an additional complexity to maintain information
consistency.
This work proposes a replicated multiversion
file server that aims to provide high data availability and operation
atomicity. The file server explores the potential of distributed architectures
and, at the same time, avoids that inconsistencies make system utilization more
difficult. In order to measure the complexity in system design, a prototype of
the distributed file server was implemented.