Failure detectors are a necessary component in many distributed applications such as business conference applications or shared distributed resource applications. These applications may rely on a failure detector component for various reasons such as security, resource availability, or performance. In a broadcast environment basic heartbeat protocols can be fast and efficient. Such a protocol becomes inefficient when used in network environments where only point-to-point communication is available since a message needs to be sent to each member in the group where only one was sufficient in a broadcast environment. This thesis investigation focused on failure detector protocols used in point-to-point communication environments. The Ping, Gossip, and Heartbeat protocols are well known failure detector algorithms. A protocol presented by Chandra, Goldszmidt, and Gupta uses a unique algorithm to confirm node failures resulting in increased accuracy while decreasing messages sent. Their protocol suffers in other areas due to the randomness of the protocol. Part of the thesis was the development of a new protocol based on improvements of Chandra, Goldszmidt, and Gupta's protocol. Each of the protocols were analyzed in detail and comparisons were made in terms of message usage, accuracy, speed, and complexity. The five protocols were implemented as well as a highly flexible testing environment. Experiments were conducted to compare the protocols and conclusions were drawn as to what protocols are best suited for different application demands and network conditions.
Library of Congress Subject Headings
Computer networks--Security measures; Computers--Reliability; Computer network protocols
Department, Program, or Center
Computer Science (GCCIS)
Dahlgren, Jeremy, "Efficient failure detection for point-to-point communication networks" (2006). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus