In the age of growing demand for data, there is also a parallel demand for the data to be provided instantaneously. Quick access to information has become a paramount pillar of the information age. As a result, system architects have invented new methods of speeding up their servers to handle requests faster than ever before. As hardware itself also quickens, some engineers turned to another source of potential slowdown, the operating system (OS), to cut corners. The creation of the Berkeley Packet Filter and, later, the Extended Berkeley Packet Filter (eBPF, though often shortened to just BPF) allow for just that. Combined with the eXpress Data Path (XDP) toolset, packet latency and throughput from client to server achieve better rates than relying upon the kernel to handle packets in many scenarios. This thesis aims to explore avenues of performance benefit and the potential applications and limitations of eBPF.
