A QoS-configurable failure detection service for internet applications

Abstract

Unreliable failure detectors are a basic building block of reliable distributed systems. Failure detectors are used to monitor processes of any application and provide process state information. This work presents an Internet Failure Detector Service (IFDS) for processes running in the Internet on multiple autonomous systems. The failure detection service is adaptive, and can be easily integrated into applications that require configurable QoS guarantees. The service is based on monitors which are capable of providing global process state information through a SNMP MIB. Monitors at different networks communicate across the Internet using Web Services. The system was implemented and evaluated for monitored processes running both on single LAN and on PlanetLab. Experimental results are presented, showing the performance of the detector, in particular the advantages of using the self-tuning strategies to address the requirements of multiple concurrent applications running on a dynamic environment.

Data-Driven Packet Loss Estimation for Node Healthy Sensing in Decentralized Cluster

“…By comparing the current time and a predicted next receipt time calculated based on these two windows, the monitor decides whether the remote node is reachable or not. Turchetti et al [12] proposes an IFDS framework which can handle multiple concurrent applications with different QoS requirements, whose purpose is different from us.…”

“…Existing failure detection methods [5,6,7,8,9,10,11,12] are designed to detect completely unreachable nodes, e.g., died or disconnected. A commonly used failure detection approach in a cluster is that, monitors estimate the state of each node based on the $\varphi$ FD each node in the cluster is monitored by a set of other nodes.…”

“…However, there are two main drawbacks of this kind of implementation. First, the existing failure detectors [5,6,7,8,9,10,11,12] cannot correctly detect a partially working node, e.g., a node which randomly drops packets. In this case, different monitors may give out different detection results on this node.…”

“…As the $\varphi$ FD and other failure detection mechanisms proposed in [5,6,7,8,9,10,11,12] are not suitable for the scenario of stochastic packet loss, we propose an algorithm that can estimate the severity of packet loss of a link between two nodes based on the statistical information of TCP protocol, the round-trip-time (RTT). Further, we propose a model that can sense a node’s healthy status from anywhere of the cluster without the limitation of any node to be always reachable, choose a unique leader without election process and make more reliable decisions about removing faulty nodes.…”

NFV‐FD: Implementation of a failure detector using network virtualization technology

“…Failure detectors (FDs) are a basic distributed system abstraction that is used to monitor processes, ie, programs running on multiple hosts of a network . Given a set of processes that can execute a distributed application, an FD is used to determined which of those processes are fault‐free and which are suspected to have crashed . Failure detectors are said to be inherently unreliable, as they were proposed for asynchronous systems in which it is impossible to know the upper limit on the delay for transmitting a message or executing a task.…”