Synchronous Distributed Load Balancing on Totally Dynamic Networks

Bahi, Jacques M.; Couturier, Raphaël; Vernier, Flavien

doi:10.1109/ipdps.2007.370572

“…Figure. 4 shows that the response time improved when the number of nodes was increased. However, this improvement was mainly caused by the fact that more nodes were used for larger domain.…”

Section: Response Timementioning

confidence: 98%

Untitled

2019

IJDPS

0

View full text Add to dashboard Cite

Load-balancing techniques have become a critical function in cloud storage systems that consist of complex heterogeneous networks of nodes with different capacities. However, the convergence rate of any load-balancing algorithm as well as its performance deteriorated as the number of nodes in the system, the diameter of the network and the communication overhead increased. Therefore, this paper presents an approach aims at scaling the system out not up -in other words, allowing the system to be expanded by adding more nodes without the need to increase the power of each node while at the same time increasing the overall performance of the system. Also, our proposal aims at improving the performance by not only considering the parameters that will affect the algorithm performance but also simplifying the structure of the network that will execute the algorithm. Our proposal was evaluated through mathematical analysis as well as computer simulations, and it was compared with the centralized approach and the original diffusion technique. Results show that our solution outperforms them in terms of throughput and response time. Finally, we proved that our proposal converges to the state of equilibrium where the loads in all in-domain nodes are the same since each node receives an amount of load proportional to its capacity. Therefore, we conclude that this approach would have an advantage of being fair, simple and no node is privileged.

show abstract

“…Bahi et al [2][3][4] and Elsässer et al [18] were the first to consider load balancing in dynamic networks. These works, alongside with a more application-oriented work of Sider and Couturier [35], were the base for the work of Dinitz et al [14], upon which we elaborate.…”

Section: Related Workmentioning

confidence: 99%

On the Complexity of Load Balancing in Dynamic Networks

Gilbert¹,

Meir²,

Paz³

et al. 2021

Preprint

0

View full text Add to dashboard Cite

In the load balancing problem, each node in a network is assigned a load, and the goal is to equally distribute the loads among the nodes, by preforming local load exchanges. While load balancing was extensively studied in static networks, only recently a load balancing algorithm for dynamic networks with a bounded convergence time was presented. In this paper, we further study the time complexity of load balancing in the context of dynamic networks.First, we show that randomness is not necessary, and present a deterministic algorithm which slightly improves the running time of the previous algorithm, at the price of not being matching-based. Then, we consider integral loads, i.e., loads that cannot be split indefinitely, and prove that no matching-based algorithm can have a bounded convergence time for this case.To circumvent both this impossibility result, and a known one for the non-integral case, we apply the method of smoothed analysis, where random perturbations are made over the worst-case choices of network topologies. We show both impossibility results do not hold under this kind of analysis, suggesting that load-balancing in real world systems might be faster than the lower bounds suggest.

show abstract

“…Presented here is a summary of work related to the approaches and techniques used in this paper [2,4,5,6,7]. In complex huge systems, it is not feasible for a node to collect the information of all other nodes in the system.…”

Section: Related Workmentioning

confidence: 99%

Advanced Diffusion Approach to Dynamic Load-Balancing for Cloud Storage

Daraghmi¹,

Daraghmi²

2019

IJDPS

View full text Add to dashboard Cite

Load-balancing techniques have become a critical function in cloud storage systems that consist of complex heterogeneous networks of nodes with different capacities. However, the convergence rate of any load-balancing algorithm as well as its performance deteriorated as the number of nodes in the system, the diameter of the network and the communication overhead increased. Therefore, this paper presents an approach aims at scaling the system out not up -in other words, allowing the system to be expanded by adding more nodes without the need to increase the power of each node while at the same time increasing the overall performance of the system. Also, our proposal aims at improving the performance by not only considering the parameters that will affect the algorithm performance but also simplifying the structure of the network that will execute the algorithm. Our proposal was evaluated through mathematical analysis as well as computer simulations, and it was compared with the centralized approach and the original diffusion technique. Results show that our solution outperforms them in terms of throughput and response time. Finally, we proved that our proposal converges to the state of equilibrium where the loads in all in-domain nodes are the same since each node receives an amount of load proportional to its capacity. Therefore, we conclude that this approach would have an advantage of being fair, simple and no node is privileged.

show abstract

Synchronous Distributed Load Balancing on Totally Dynamic Networks

Abstract: Abstract

Cited by 8 publications

References 13 publications

Untitled

Untitled

On the Complexity of Load Balancing in Dynamic Networks

Advanced Diffusion Approach to Dynamic Load-Balancing for Cloud Storage

Contact Info

Product

Resources

About