MTU Analyzing for Data Centers Interconnected Using VxLAN

Abstract

Network virtualization offers advanced technology solutions for data centers and clouds, specifically through the use of Virtual eXtensible Local Area Network (VxLAN) to interconnect multiple data centers. Nonetheless, it is crucial to take into account the Maximum Transmission Unit (MTU) when implementing overlay VxLAN technology. This technology introduces additional overhead to network packets, and exceeding the MTU can result in performance degradation and increased processing overhead due to packet fragmentation. The Virtual Tunneling End Point (VTEP) is employed in VxLAN overlay tunneling but does not forward packets that exceed the path MTU, leading to packet loss and significant delays caused by network misconfiguration. This paper thoroughly examines the issues related to MTU size and emphasizes the overhead introduced by overlay technologies like VxLAN. Additionally, it highlights the necessity of implementing MTU discovery features. Simulation results demonstrate that enabling MTU discovery features allows hosts to avoid packet loss while ensuring network stability and adaptability. Conversely, manually adjusting the MTU size may be restricted or blocked by device vendors, especially when data centers are interconnected via third-party networks. By comparing the manual configuration approach with MTU discovery methods, data center administrators can make informed decisions. The simulation results clearly indicate that MTU discovery surpasses manual configuration in terms of throughput, reduces packet loss, and minimizes delays.