Increased IP layer protection bandwidth due to router bypass

Abstract

Counter to previous works on router bypass, that assume protection bandwidth in routers is independent of the amount of bypass, we show that more bypass implies more IP protection bandwidth to cope with reduced likelihood of sharing. IntroductionMany studies (e.g., [1]) have investigated the cost savings that can be achieved if routers are bypassed by either optical switches (e.g., ROADMs) or electrical switches (SONET/SDH and more recently OTN). The design methodology in these studies is typically simple: for each demand between routers, find if it can bypass intermediate routers in the transport layer, either individually -over a dedicated wavelength, or sharing a wavelength with additional demands using TDM grooming. If the demand cannot adequately fill the transport container (OTN connection or wavelength depending on the case), then it is aggregated with other IP demands using routers, until the aggregate can fill up the container and bypass some routers. Note that if the design does not include OTN, the resulting IP layer design will be sparser (i.e., includes fewer links) than a design that includes OTN, since there are less opportunities to map traffic efficiently into wavelengths than opportunities to map it into (a smaller) OTN container. The finer the granularity of OTN, the denser the IP network will become.