Gossip

If you only read one of these papers, start with the classic Demers, et al paper:

• Epidemic Algorithms for Replicated Database Maintenance

Peer sampling services

Briefly, a peer sampling service is a system that maintains a restricted set (partial view) of the all machines participating in a gossip system.

• The Peer Sampling Service: Experimental Evaluation of Unstructured Gossip-Based Implementations
• HyParView: a membership protocol for reliable gossip-based broadcast
• SWIM: Scalable Weakly-consistent Infection-style Process Group Membership Protocol

Epidemic broadcast

• Large-Scale Newscast Computing on the Internet
• Bimodal Multicast
• Efficient Reconciliation and Flow Control for Anti-Entropy Protocols
• Epidemic Broadcast Trees

Failure Detectors

• A Gossip-Style Failure Detection Service
• The ϕ Accrual Failure Detector