Comparative study for control plane scalability approaches in SDN productive networks
Mandisa N. Nxumalo, Ijeoma Mba, Matthew Olusegun Adigun · 2017
The high increase in technology invasion leads to a tremendous increase in its use and demand. This contributes to high demand and usage of productive networks such as data centers, campus networks, cloud and enterprise networks. However, these productive networks require a high computing power in order to efficiently handle high data flows as the network increases. In SDN, an increase in the number of tenants in a network does not scale the centralized controller and can lead to lower production, performance and reliability levels. Approaches such as Kandoo, Onix and HyperFlow are used to enable scalability on a control plane of different SDN productive network environments. This work addresses the scalability issue through carrying out a comparative study on some highly rated approaches based on the literature review in order to select an optimal approach for control plane scalability. The Kandoo and Onix approach possessed the optimal conditions given for an optimal scalability approach. Due to the openness issue, only the Kandoo approach was emulated based on a similar experiment done on Onix. The performance indicated that both approaches are individually outstanding in their performance in accordance to the Literature. However, emulation results show that Kandoo depicts less complexity in its implementation compared with Onix. The Kandoo approach also supports both local and rare events as against Onix, which, though supports a large-scale network than Kandoo but proved to be a daunting task in implementation and does not support local events.