SDLB: A Scalable and Dynamic Software Load Balancer for Fog and Mobile Edge Computing
SDLB: A Scalable and Dynamic Software Load Balancer for Fog and Mobile Edge Computing
复制标题
DOI：
10.1145/3098208.3098218
复制
发表时间：
2017-08
期刊：
影响因子：
--
通讯作者：
Ye Yu;Xin Li;Chen Qian
中科院分区：
文献类型：
--
作者：
Ye Yu;Xin Li;Chen Qian
移动边缘计算（MEC）向网络边缘的移动设备提供了计算/存储卸载和资源虚拟化。负载平衡器是从移动设备确定每个数据包的目标MEC主机的必要网络功能，以进行这种虚拟化。由于MEC的新特征，例如资源限制和高动力学，因此无法将云负载平衡器的现有解决方案直接应用于MEC。本文介绍了满足MEC要求的可扩展和动态负载平衡器的新设计。 SDLB的核心算法是最小的完美哈希，它提供了两个完美的功能作为负载平衡器。评估结果表明，SDLB比云的广泛使用负载平衡器设计要快4倍至10倍，并且使用（<50％）的内存。
Mobile Edge Computing (MEC) provides computing/storage offloading and resource virtualization to mobile devices at the network edge. A load balancer is a necessary network function to determine the destination MEC host of each packet from a mobile device, for such virtualization. Due to the new characteristics of MEC, such as resource limitation and high dynamics, existing solutions of cloud load balancer cannot be directly applied to MEC. This paper presents a new design of a Scalable and Dynamic Load Balancer, called SDLB, that satisfies the requirements of MEC. The core algorithm of SDLB is minimal perfect hashing, which provides two perfect features as a load balancer. Evaluation results show that SDLB is faster by 4x to 10x and uses much less (< 50%) memory, than a widely-used load balancer design for cloud.