Bandwidth Guarantees for Resilient Ethernet Networks through RSTP Port Cost Optimization Andr´ as Kern, Istv´ an Moldov´ an and Tibor Cinkler Department of Telecommunications and Media Informatics Budapest University of Technology and Economics Magyar tud´ osok krt. 2, Budapest, Hungary H-1117 email: {kern, moldovan, cinkler}@tmit.bme.hu Abstract— Ethernet protocol is the most wide-spread protocol in the LAN environment. It is cost effective, simple, and provides high speeds, exactly what is needed in the provider network. However, deployment in the provider network imposes carrier- grade requirements. Standardization bodies realized this, and they are extending its capabilities: QoS support and traffic man- agement by 802.1Q VLANs, traffic engineering by 802.1s MSTP, OAM by 802.1ag CFM. However, with carrier-grade extensions Ethernet started to lose an important property: simplicity. Often the topology of the Ethernet aggregation is simple, tree-like, so complex protocols like MSTP are not required even when resilience is supported by adding several links. In most cases RSTP can provide the necessary restoration and loop protection capability. However, the default settings of the RSTP protocol may not provide optimal network utilization, and it is difficult to predict its behavior after a failure. In this paper we present a tool for RSTP optimization, which engineers the network for best utilization while also ensures that optimal paths will be selected after a link failure. The result of the optimization is a port cost set for all bridge interfaces. The optimization is performed off- line using an ILP formulation. We show that compared to the default settings higher throughput can be achieved, while the required bandwidth is guaranteed in case of any link failure. I. I NTRODUCTION In the next generation networks there is a clear convergence to a multi-service network offering more than best effort internet access, but also voice and video services over the same infrastructure. These new services require QoS guarantees and high avail- ability, besides the increased bandwidth. For instance, Voice over IP (VoIP) service expects very low delay and jitter, while the video streaming services are sensitive to packet loss too. At the same time, customers have very high expectations on these services: they want to receive the same quality and availability as provided by the traditional operators over their dedicated infrastructure. Therefore, restoration times in order of seconds are anticipated, demanding resilient networks. In the Metropolitan area Ethernet drives the convergence in triple play networks. Ethernet provides very high speeds with copper or low-cost optical interfaces and its management is simple. Also, Ethernet is the predominant interface on the endpoints that need to be connected. Most of the Ethernet equipment supports the spanning tree protocols those are responsible for path selection and restoration while maintain- ing a loop-free logical topology. The Rapid Spanning Tree protocol (RSTP) [1] provides low restoration times (ranging from milliseconds to several second depending on topology), while it remains simple from the management point of view. Traffic separation and QoS are provided by the Virtual LANs (VLANs) defined by IEEE802.1Q [2]. The standard adds a VLAN tag to Ethernet frames that allows 4096 VLANs and 3 priority bits that makes possible traffic class differenti- ation. Resilience in Metro Ethernet network can be realized by Link Aggregation (IEEE 802.3ad [3]), or by deploying dual- homing and rings topologies. Link Aggregation bundles links together provide increased bandwidth during normal operation and it protects links in the network using different physical paths for links to ensure protection. In the same time, it still requires the use of a spanning tree protocol in the network making the Link Aggregation being an auxiliary solution for protection only. On the other hand, RSTP supports restoration by automat- ically reconstructing the forwarding topology after a failure and it can work over arbitrary topology. However, RSTP with the default cost settings (which is based on the link speeds [2]) may not achieve good network utilization, moreover in case of failure it is difficult to predict the throughput on the restored topology. In this paper we present an optimization tool, which not only optimizes the forwarding topology, but also optimizes the tree for each link failure. This way bandwidth guarantees are given even in case of a link failure. II. TECHNICAL BACKGROUND The simple frame forwarding scheme of Ethernet works well only with trees, but Ethernet should work over arbitrary topology in order to realize resilience and Traffic Engineering (TE). The Spanning-Tree Protocol (STP), which is proposed in initial version of IEEE 802.1D [4], is developed to address this problem. The STP is responsible for building a loop-free logical forwarding topology over the physical one providing connectivity among all nodes. The links that are not part of this tree are blocked. In case of a failure, the blocked links are activated providing a self-healing restoration mechanism. All information propagated between the switches is embedded in Bridge Protocol Data Units (BPDUs). These packets are exchanged only between adjacent bridges, and protocol events