The integrated services packet network (ISPN) architecture proposed within the Internet community incorporates a resource reservation mechanism for those applications requiring quality of service (QoS) guarantees. Resource reservation introduces a new form of resource contention that can lead to reduced network throughput and thrashing. We establish several necessary conditions to induce thrashing. We also look at the effects several different reservation models and user behavior can have on network stability. Our work is unique from previous network resource reservation investigations in that we consider the effects of reservations for multipoint-to-multipoint applications. We conclude with examples of how simple modifications to user behavior can result in significant increases in system stability.

This paper presents a comparative analysis of two resource reservation protocols, ST-II proposed by Topolcic (1990) and resource reservation protocol (RSVP) proposed by Zhang, Braden, Estrin, Herzog and Jamin (1994) in support of an integrated services packet network (ISPN). The authors use simulations to examine the network-wide resource requirements for each protocol to support a number of application communication styles, across a range of group sizes and membership distributions. They also present a comparison of the protocol features to accommodate network and group membership dynamics.>

The goal of network design is to meet the needs of resident applications in an efficient manner. Adding real-time service and point-to-multipoint multicast routing to the Internet's traditional point-to-point best effort service model will greatly increase the Internet's efficiency in handling point-to-multipoint real-time applications. Recently, the RSVP resource reservation protocol has introduced the concept of “reservation styles”, which control how reservations are aggregated in multipoint-to-multipoint real-time applications. In this paper, which is an extension of [9], we analytically evaluate the efficiency gains offered by this new paradigm on three simple network topologies: linear, m-tree, and star. We compare the resource utilization of more traditional reservation approaches to the RSVP reservation styles in the asymptotic limit of large multipoint applications. We find that in several cases the efficiency improvements scale linearly in the number of hosts.

The current internet is based on a stateless (datagram) architecture. However, many recent proposals rely on the maintenance of state information within network routers, leading to the authors' interest in the implications of a stateful network layer. They collected internetwork traffic traces at the border routers of stub and transit networks, and used these data to evaluate, or predict, the effects of design alternatives for stateful architectures. They present an estimate of the number of active conversations at a router, and from this derive the storage requirements for the associated conversation state table. The analysis shows that, at the network periphery, fine-grain control over the traffic may be possible. However, deeper within the network, it may be more efficient to manage the conversations at a coarser level. The network traffic traces are used to perform trace driven simulations of an LRU cache, for different conversation granularities. Results show that locality exists for each of the conversation types investigated.>

In this paper, we characterize wide-area network applications that use the TCP transport protocol.We also describe a new way to model the wide-area traffic generated by a stub network.We believe the fmffic model presented here will be useful in studying congestion control, routing algorithms, and other resource management schemes for existing and future networks.Our model is based on trace analysis of TCP/IP widearea intemetwork traffic.We collected the TCP/IP packet headers of USC, UCB, and Bellcore networks at the point they connect with their respective regional access networks.We then wrote a handful of programs to analyze the traces.Our model characterizes individual TCP conversations by the distributions ofi number of bytes transfemed, duration, number of packets transferred, packet size, and packet interarrival time.Our trace analysis shows that both interactive and bulk transfer traffic from all sites reflect a large number of short conversations.Similarly, it shows that a very large percentage of traffic is bidirectional, even for bulk transfer.We observed that interactive applications send significantly different amounts of data in each direction of a conversation, and that interarrival times for interactive applications closely follow a constant plus exponential model.Half of the conversations are directed to a handful of networks, but the other half are directed to hundreds of networks.Many of these observations contradict commonly held beliefs regarding wide-area traffic.