Towards improved load balancing for data intensive distributed computing

Author(s):  
Sven Groot ◽  
Kazuo Goda ◽  
Masaru Kitsuregawa
2021 ◽  
Author(s):  
Teodor Malbašić ◽  
Petar D. Bojović ◽  
Živko Bojović ◽  
Jelena Šuh ◽  
Dušan Vujošević

Abstract Software-defined networking (SDN) provides many benefits, including traffic programmability, agility, and network automation. However, budget constraints burdened with technical (e.g., scalability, fault tolerance, security issues) and, sometimes, business challenges (user acceptance and confidence of network operators) make providers indecisive for full SDN deployment. Therefore, incremental deployment of SDN functionality through the placement of a limited set of SDN devices among traditional devices represents a rational and efficient environment that can offer customers modern and more data-intensive services. However, while hybrid SDN provides many benefits, it also has specific challenges addressed in the literature. This paper answers one of these challenges by presenting the research and development of a new load balancing scheme in the hybrid SDN environment built with a minimal SDN device set (controller and one switch). We propose a novel load balancing scheme to monitor current server load indicators and apply multi-parameter metrics for scheduling connections to balance the load on the servers as efficiently as possible. The base of the new load balancing scheme is continuous monitoring of server load indicators and implementations of multi-parameter metrics (CPU load, I/O Read, I/O Write, Link Upload, Link Download) for scheduling connections. The testing performed on servers aims to balance the server's load as efficiently as possible. The obtained results have shown that this mechanism achieves better results than existing load balancing schemes in traditional and SDN networks. Moreover, a proposed load balancing scheme can be used with various services and applied in any client-server environment.


2014 ◽  
Vol 7 (1) ◽  
pp. 267-281 ◽  
Author(s):  
B. van Werkhoven ◽  
J. Maassen ◽  
M. Kliphuis ◽  
H. A. Dijkstra ◽  
S. E. Brunnabend ◽  
...  

Abstract. The Parallel Ocean Program (POP) is used in many strongly eddying ocean circulation simulations. Ideally it would be desirable to be able to do thousand-year-long simulations, but the current performance of POP prohibits these types of simulations. In this work, using a new distributed computing approach, two methods to improve the performance of POP are presented. The first is a block-partitioning scheme for the optimization of the load balancing of POP such that it can be run efficiently in a multi-platform setting. The second is the implementation of part of the POP model code on graphics processing units (GPUs). We show that the combination of both innovations also leads to a substantial performance increase when running POP simultaneously over multiple computational platforms.


Sign in / Sign up

Export Citation Format

Share Document