Learning from the past for resolving dilemmas of asynchrony

Authors:
Paul Ezhilchelvan;Santosh Shrivastava
Affiliations:
Newcastle University, UK;Newcastle University, UK
Venue:
ACM SIGOPS Operating Systems Review
Year:
2010

Citing 13
Cited 1

Congestion avoidance and control

SIGCOMM '88 Symposium proceedings on Communications architectures and protocols
Impossibility of distributed consensus with one faulty process

Journal of the ACM (JACM)
Unreliable failure detectors for reliable distributed systems

Journal of the ACM (JACM)
Implementing Fail-Silent Nodes for Distributed Systems

IEEE Transactions on Computers
The Timed Asynchronous Distributed System Model

IEEE Transactions on Parallel and Distributed Systems
Byzantine generals in action: implementing fail-stop processors

ACM Transactions on Computer Systems (TOCS)
Real-Time Systems: Design Principles for Distributed Embedded Applications

Real-Time Systems: Design Principles for Distributed Embedded Applications
Practical byzantine fault tolerance and proactive recovery

ACM Transactions on Computer Systems (TOCS)
The Timely Computing Base Model and Architecture

IEEE Transactions on Computers
Design and Evaluation of a QoS-Adaptive System for Reliable Multicasting

SRDS '04 Proceedings of the 23rd IEEE International Symposium on Reliable Distributed Systems
A Performance Study on the Signal-On-Fail Approach to Imposing Total Order in the Streets of Byzantium

DSN '06 Proceedings of the International Conference on Dependable Systems and Networks
The Chubby lock service for loosely-coupled distributed systems

OSDI '06 Proceedings of the 7th symposium on Operating systems design and implementation
No time for asynchrony

HotOS'09 Proceedings of the 12th conference on Hot topics in operating systems

Predictable service overlay networks: Predictability through adaptive monitoring and efficient overlay construction and management

Journal of Parallel and Distributed Computing

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper presents two design approaches to avoid many complications introduced at both user and developer levels by the FLP impossibility. The first approach is appropriate in managed hosting environments, such as datacenters, and involves offering service guarantees with tunable success probabilities and remedial actions in the unlikely scenarios. The second is appropriate in open environments and advocates building fail-signal abstractions for hosting application-level replication.