Americas

  • United States
by Tejas Vashi, special to Network World

RPR ups efficiency of metro Ethernet

How-To
May 31, 20043 mins
Networking

Companies traditionally have gained basic resiliency for their metropolitan Ethernet networks through a Layer 2 mechanism such as Spanning Tree or Layer 3 routing protocols. But these mechanisms, with their few-second network restoration times, are inadequate for delay- and jitter-sensitive data applications that are proliferating rapidly on enterprise networks. To gain greater resiliency, corporations have had to install fixed-bandwidth point-to-point circuits, or fiber pairs, between campus nodes along with back-up paths for protection – a costly endeavor.

What companies need is a metropolitan Ethernet architecture with the resiliency of SONET but with significant network efficiencies for packet-based services. Resilient Packet Ring (RPR) is an emerging Layer 2 media access control (MAC) technology that meets those needs. The IEEE created a draft specification for RPR and is expected to ratify the standard this year.

RPR uses Ethernet switching and a dual counter-rotating ring topology to provide SONET-like network resiliency and optimized bandwidth usage, while delivering multipoint Ethernet/IP services. RPR maintains its own protection scheme and uses physical-layer alarm information and Layer 2 protocol communications to detect node and/or link failures. When a failure is detected, the RPR switching mechanism restores networks in less than 50 millisec.

Because RPR is a Layer 2 MAC-based technology, it can operate over multiple physical layers, including SONET. Therefore, corporations can reap the benefits of RPR by having it ride over the SONET network to deliver the resilient, efficient, multipoint functionality and scalability of data applications such as VoIP, packet video, business continuance and distance learning.

Or they can install multiservice provisioning platforms, which are optimized for TDM services but also can support advanced data applications via RPR over SONET. The advantage is that existing TDM services are maintained, while a smooth migration to packet-based services is enabled.

Another major advantage of RPR’s dual-rotating ring design is that Ethernet traffic is sent in both directions on the ring to achieve the maximum bandwidth utilization. Unlike older ring-based data networks such as token ring or FDDI, RPR uses a spatial reuse mechanism. Rather than requiring traffic to traverse the entire ring even though a destination node is only a hop away, RPR sends it there directly, keeping the rest of the ring bandwidth available for use by other stations on the network.

To further enhance the network efficiency and support multimedia applications, the IEEE has included a classification scheme and a fairness algorithm in the RPR specification. This guarantees that jitter- and delay-sensitive traffic is always given higher-priority access to the network. Meanwhile, best effort (Internet type) data traffic is ensured equal access and a “fair” share of the remaining ring bandwidth.

RPR also uses statistical multiplexing so that bandwidth can be oversubscribed, while establishing committed information rate (CIR) and peak-rate thresholds on a per-application basis. This guarantees each enterprise application a CIR and the ability to burst up to the peak rates when bandwidth is available. With such a mechanism, each department is charged only for using extra bandwidth rather than being billed for a larger, nailed-up circuit, regardless of use.

Widespread corporate adoption of RPR will help usher in the cost-effective transport of popular Ethernet and IP communications services. RPR transport will provide efficient bandwidth protection, accommodate bursty data traffic and provide the quality of service needed for these advanced packet applications.

Vashi is a senior product line manager at Cisco. He can be reached at tvashi@cisco.com.