An Improved Switch Migration Decision Algorithm for SDN Load Balancing

Abstract

Dynamic and Adaptive Load Balancing (DALB) and Controller Adaption and Migration Decision (CAMD) frameworks are the recently developed efficient controller selection frameworks that solved the challenge of load-imbalance in Software-Defined Networking (SDN). While CAMD framework was established to be efficient over DALB framework yet it was not efficient when the incoming-traffic load was elephant flow, hence, leading to a significant reduction in the overall system performance. This study had proposed an Improved Switch Migration Decision Algorithm (ISMDA) that solved the network challenge when the incoming load is elephant flow. The balancing module of the switch migration framework, which runs on each controller, is initiated during the controller load imbalance phase. The improved framework used the controller variance and controller average load status to determine the set of underloaded controllers in the network. The constructed efficient migration model was used to, simultaneously, identify both the migration cost and load-balancing variation for the optimal selection of controller among the set of underloaded controllers. The controller throughput, response time, number of migration space and packet loss were used as the performance comparison metrics. The average controller throughput of ISMDA increased with 7.4% over CAMD framework while average response time of the proposed algorithm improved over CAMD framework with 5.7%. Similarly, the proposed framework had 5.6% average improved migration space over CAMD framework and the packet-loss of ISMDA had average 6.4% performance over the CAMD framework. It was concluded that ISMDA was efficient over CAMD framework when the incoming traffic load is elephant flow.

Evaluation Test and Improvement of Load Balancing Algorithms of Nginx

“…Then the maximal weight (CWeights[Maindex]) will subtract the total weight(lines 10), and we can get a new list called effective weights: 2, 4, and -6. Finally, we can get the sum of each weight of CWeights[ ] and EWeights[ ]: 4, 8, and 0(lines [11][12], and this algorithm will continue the same loop with a new CWeights[ ] one by one(lines 4).…”

“…Secondly, the w is the remainder of hash % total_weight(lines 5), and the peer is pointed to the beginning of the array EWeights[ ](lines 6). Finally, the algorithm traverses back-end nodes, subtracts the weight of each node by w in turn, and calculates the index of the backend node where w is less than its weight (lines [8][9][10][11][12]. This time, the SeNode is the selected node, and then the algorithm will continue the next loop.…”

“…The algorithm results are analyzed from two aspects, which makes the conclusion more convincing. In reference [12], this paper created an improved switch migration decision algorithm (ISMDA) that solved the network challenge when the load is elephant flow. The performance comparison metrics include the controller throughout, response time, number of migration space, and packet loss, and this algorithm constructed a relatively complete measurement model.…”

Multi-Controller Load Balancing Algorithm for Test Network Based on IACO

“…(3) When a load imbalance occurs, the switch is migrated from an overloaded controller to an underloaded one. Thus, load balancing is achieved by dynamically changing the relationship between the switch and the controller [23][24][25].…”

“…However, the DDM used a greedy algorithm to select the target controller so that the results may not be the global optimal solution. [24] proposed an improved switch migration decision algorithm (ISMDA). The improved algorithm uses the controller variance and the average load state of the controller to determine the set of underloaded controllers in the network, but the migration cost is not considered to be enough.…”