Introduction

The main goal of this implementation is to make a networking device that can schedule IP traffic on existing network interfaces. Currently scheduling of packets is conducted according to a scheme known as Weighted Fair Queuing [2].

The device understands the RSVP protocol [3], interprets it and uses information from the protocol to set parameters in the WFQ scheduler.

In order to make clear which terms refer to the architecture the following diagram should be used as reference

                RSVP (or protocol)                      
                                                        user
                ------------------------------------
                                                        kernel
                Scheduler (or Traffic Control Module)

In this text we use RSVP and protocol interchangeably, likewise so for scheduler and Traffic control module. The boundary between the protocol and the scheduler is also evident in their separation across user and kernel spaces respectively. Interactions between the two modules is discussed later.

An important secondary goal is to without any modifications to the device driver. Other queuing schedulers [1] have been written but require modifications to the driver or the IP stack in some way.

This report will start with a basic introduction to the system demonstrated with a simple 3 entity system, 2 end systems and a single router. In subsequent sections the internals of the scheduler are explained followed by some details about the protocol. Some examples of how to use the system are given followed by how to extend/change and if need be debug the system.