An Experimental review on Intel DPDK L2 Forwarding

摘要

Packet processing using the traditional techniques can be very time-consuming and resource usage for the same is very inefficient. The issue of slow processing arises from the fact that the processing techniques need to interact with the Kernel layer quite intimately and this is a laborious task. Intel DPDK is an integrated set of libraries and network interface controller drivers which can lead to a paradigm shift from using discrete working components for each module of packet processing to a consolidated model, which is simplified and shows scalability. The DPDK has a set of data plane libraries that can improve packet processing performance by many times the current processing strength provided by a traditional OS. The idea is to allow users to integrate their application software with DPDK libraries to access and make use of packet transfer acceleration provided by DPDK. This paper explains the basic architecture and functioning of DPDK along with one of its simple applications-The L2 forwarding process.[1] The results of the experimental application show the stable increment to performance that DPDK has provided. The analysis done on L2 forwarding application deduces the boost that DPDK libraries give to packet processing.