Network Coding to Reduce Congestion and Improve Memory Buffer in Smart Switch
Souryendu Das, Stavros Kalafatis · 2023
Content Distribution Networks (CDNs) are gaining popularity with an ever increasing demand of streaming services. As enterprises manage larger amounts of data, they are continuously researching improvements for their data center networks. Various algorithms have been proposed to diligently manage these needs. This paper proposes a new algorithm called NCSSDDA (Network Coded SSSDDA) which incorporates network coding to improve the performance of an existing algorithm called SSDDA (Smart Switch Dynamic Delay Algorithm) in a data center network. The NCSSDDA algorithm aims to reduce packet drop rates and memory consumption of switch memory buffer, with a slight tradeoff of reduced line rate. Experiments were conducted in a data center rack with multiple endpoint servers and results show that the NCSSDDA algorithm provides algorithmic benefits over the original SSDDA.