A Simulationstudyofipswitceiing

Recently there has been much interest in combining the speed of layer-2 switching with the features of layer-3 routing. This has been prompted by numerous proposals, including: IP Switching [l], Tag Switching [2], ARIS [3], CSR [4], and IP over ATM [5]. In this paper, we study IP Switching and evaluate the performance claims made by Newman et al in [l] and [6]. In particular, using ten network traces, we study how well IP Switching performs with traffic found in campus, corporate, and Internet Service Provider (ISP) environments. Our main finding is that IP Switching will lead to a high proportion of datagrams that are switched; over 75% in all of the environments we studied. We also investigate the effects that different flow classifiers and various timer values have on performance, and note that some choices can result in a large VC space requirement. Finally, we present recommendations for the flow classifier and timer values, as a function of the VC space of the switch and the network environment being served. What is IP Switching? The Internet is overwhelmed with traffic, with user demand for bandwidth well in excess of supply. This should come as no great surprise the number of users of the Internet has grown exponentially for ten straight years, with each user adding more demand to the network. Over time, each user’s traffic tends to increase, adding further to the demand. And at the same time, a larger proportion of traffic traverses multiple domains, swamping campus backbones, corporate intranets, and the core of the Internet. In response to the demand, network equipment vendors and service providers have developed and deployed faster and faster routers. Yet demand for bandwidth far outstrips the current link and switching capacity of the Internet. The performance of network routers is usually limited by two main tasks: (1) Examining the destination address of incoming datagrams, indexing into a routing table to determine the next hop, and (2) copying incoming datagrams to their outgoing interface. The routing lookup is not straightforward; the router must find the longest prefix match over a space of approximately 231 possible addresses, It is unusual for a router today to perform more than 100,000 address lookups per second. Perhaps surprisingly, the performance of most current routers is limited by their ability to copy data between interfaces. This is because datagrams frequently traverse shared busses multiple times before reaching their outgoing interface. A variety of techniques have been proposed to accelerate routing lookups [7], and replace shared busses with switched backWork supported by the Center for Integrated Systems at Stanford University. PermIssIon to moke digital/hard copy of part or all this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantoge, the copyright notice, the title of the publication and its date appear, and notice is given that copying is by permission of ACM, fno, To copy otherwise, to republish, to post on servers, or to redfstrlbute to J/sts, requires prior specific permission and/or a fee. SIGCOMM ‘97 Cannes, France