TABLE OF CONTENTS Although the focus of this book is on IP (i.e. layer 3) QOS, IP networks use underlying layer 1 and 2 technologies in order to provide connectivity between layer 3 nodes (i.e. routers). Therefore, in building end-to-end IP services with contracted SLA commitments, it is essential that the underlying layer 1 and layer 2 technologies are able to support the network requirements needed to deliver the contracted SLAs. The SLAs provided at the IP layer are, however, implicitly limited by the SLAs of the underlying layer 2 technology; for example, it would not be possible to deliver an IP service supporting VoIP with a bounded delay, jitter and loss, using an underlying layer 2 network, which did not provide SLAs for delay, jitter and loss at least as good (generally better) as those required for the IP service, e.g. an ATM ABR service.
Some point-to-point layer 2 technologies, such as leased lines delivered on underlying TDM or Synchronous Optical Network (SONET)/ Synchronous Digital Hierarchy (SDH) networks, generally have well-defined SLAs at layer 2, which are capable of supporting layer 3 applications with tight SLA requirements. Such connections generally have a defined delay, a defined bit error rate (BER) and committed layer 2 minimum bandwidth, which is normally equal to the maximum bandwidth for the service. In these cases, Intserv or Diffserv queuing policies can be attached directly to the interface that represents the connection; in the case of Diffserv the policy applied would be similar to...
