Log-Based Recovery in Asynchronous Distributed Systems
Abstract
Replication has been shown to be an important tool in the design of high-performance and highly-available distributed systems. When applied to data, however, replication significantly complicates the problem of maintaining consistency within a system. This problem is further complicated when repositories of the data can potentially fail and recover. In this dissertation, we describe a log-based mechanism for restoring consistent states to replicated data objects after failures. (rrh)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1990
- Accession Number
- ADA216833
Entities
People
- Kenneth P. Kane
Organizations
- Cornell University