Improving Raft When There Are Failures

Fluri, Christian; Melnyk, Darya; Wattenhofer, Roger

doi:10.1109/ladc.2018.00028

Cited by 9 publications

(4 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Wang 17 proposed an improved Raft algorithm, hhRaft, which is capable of supervising candidates by introducing a monitor server to identify forged Request Vote messages from malicious servers. Fluri optimized the election timeout mechanism, which minimized the probability of most servers in the cluster losing contact with the leader because of data packet loss during network transmission 18 . To improve the algorithm performance, in Reference, 19 the election timeout for servers that are isolated or with network partitions was dynamically adjusted.…”

Section: Background and Related Workmentioning

confidence: 99%

“…Fluri optimized the election timeout mechanism, which minimized the probability of most servers in the cluster losing contact with the leader because of data packet loss during network transmission 18 . To improve the algorithm performance, in Reference, 19 the election timeout for servers that are isolated or with network partitions was dynamically adjusted. Hanmer 20 modified the election timeout mechanism to dynamically change, which can avoid the increase in the heartbeat sending interval caused by the increase in the load of the leader; when the increase is greater than the election timeout of the follower, a new election round is triggered.…”

Section: Background and Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Multi‐strategy‐based leader election mechanism for the Raft algorithm

et al. 2023

Concurrency and Computation

View full text Add to dashboard Cite

SummaryRaft is a consensus algorithm that implements highly available replicas. As raft requires a leader to drive the protocol execution, it usually employs a round‐robin leader election mechanism to select a random leader among all nodes. However, when the elected leader suffers from performance issues, the protocol will temporarily lose aliveness. To address this issue, we propose a multi‐strategy leader election mechanism that allows nodes to proactively trigger a new round of election when they believe the leader suffers from performance issues, and replace this leader with a new one in a designated priority queue. The formal analysis and experimental results show that the mechanism can improve the system's throughput/latency without additional election time overhead.

show abstract

Section: Background and Related Workmentioning

confidence: 99%

Section: Background and Related Workmentioning

confidence: 99%

Multi‐strategy‐based leader election mechanism for the Raft algorithm

et al. 2023

Concurrency and Computation

View full text Add to dashboard Cite

show abstract

“…The authors argue that an overloaded leader sends messages with a high time deviation; consequently, they recommend that receivers campaign for the next leadership if they detect a time difference between the leader's messages. In [11], two algorithms to minimize the number of elections without a winner are proposed (due to the split vote, that will be further explained later).…”

Section: Related Workmentioning

confidence: 99%

When Machine Learning Meets Raft: How to Elect a Leader over a Network

Choumas,

Korakis

2023

GLOBECOM 2023 - 2023 IEEE Global Communications Conference

View full text Add to dashboard Cite

Numerous well-known applications use the Raft consensus algorithm to maintain consistent replicas of their data on distributed nodes. Raft is based on a dynamically elected leader who is one of the distributed nodes, and its operations are unfortunately suspended during the election of the leader. Elections can be triggered by the failure of the current leader, in which case they are unavoidable, or by a network disconnect between the leader and another node, in which case a new inefficient leader will likely replace the previous one at the expense of additional system downtime. In this paper, Raft messages are monitored at every node, and Machine Learning is used to classify the aforementioned causes of each election. This data is used to increase the system's availability by decreasing the total number of elections that could be conducted in a given time unit. Three supervised classifiers were trained with messages generated in a real Raft-operated distributed system that was deployed on a testbed and where multiple events triggering elections were applied. All classifiers are nearly 97% accurate at classifying the causes of these elections, approaching even 100% in some cases.

show abstract

“…Despite its relative simplicity, Raft provides a safe and efficient implementation of a state machine on top of a clustered computing system. Raft ensures reliable delivery of signals to all nodes in a given order [18]. This ensures the transition of all state machines through the same sequences of states and each node is guaranteed to come into agreement with the other nodes.…”

Section: Cluster Management Algorithmmentioning

confidence: 99%

Decentralized Management System for Solid-State Voltage Regulators in Nodes of Distribution Power Networks

Polozov¹,

Sosnina²,

Kombarov³

et al. 2021

Adv. sci. technol. eng. syst. j.

View full text Add to dashboard Cite

The article describes the concept and architecture of a decentralized control system for a solid-state voltage regulator (SSVR). The SSVR is a universal device for controlling the mode and operation parameters of medium voltage electrical networks. SSVR manage the amount of current in line using the vector voltage control method. The SSVR control system consists of two levels: the SSVR semiconductor converter control level (the technological control system), SSVR cluster management level (intelligent control system). The objective of the study is to develop an algorithm for managing the SSVR cluster located in the nodes of the distribution electric network. The Raft consensus algorithm for managing the computing cluster is applied to ensure reliable decentralized network management. The algorithm is iterative. Ethernet and PLC architectures are proposed for constructing a data transmission network between SSVR nodes. A simulation model of the SSVR cluster and its control system is developed to study the operation of the control system (node shutdown, loss of communication). The criteria for the normal operation of the intelligent control system are formulated and an algorithm for its operation in emergency situations is presented. The studies of the SSVR control system confirmed the operability of the developed control system in normal and emergency operation of the cluster on a simulation. The dependence of the control system response time on the number of cluster nodes is investigated. The maximum number of nodes in the SSVR cluster depending on the tasks being solved by the SSVR will be limited by the speed of the control system. If the number of cluster nodes increases, it is necessary to enlarge the minimum time interval between requests for control commands.

show abstract

Improving Raft When There Are Failures

Cited by 9 publications

References 2 publications

Multi‐strategy‐based leader election mechanism for the Raft algorithm

Multi‐strategy‐based leader election mechanism for the Raft algorithm

When Machine Learning Meets Raft: How to Elect a Leader over a Network

Decentralized Management System for Solid-State Voltage Regulators in Nodes of Distribution Power Networks

Contact Info

Product

Resources

About