Resilient Packet Ring basics

Resilient Packet Ring technology will likely play a pivotal role in enabling service providers to offer multipoint Ethernet services of the greatest value to enterprise customers. So we’re presenting here brief tutorial on a few things you might like to know about RPR – though, like Multi-protocol Label Switching, RPR is more relevant to the carrier’s infrastructure than to yours.

The fiber installed in major metropolitan areas was laid in a ring fashion, which was more economical for carriers than hub-and-spoke or full-mesh configurations. SONET was installed on top of the fiber primarily for handling point-to-point TDM voice, and it is circuit-based in nature.

Now, with all this fiber in the ground, why not extend the benefits of Ethernet (economies of scale, high speeds, plenty of expertise) from the LAN and onto these high-bandwidth infrastructures? The challenge has been that SONET is TDM-based and highly reliable, while Ethernet is packet-based and, well, usually requires some quality-of-service (QoS) assistance.

SONET has built-in resiliency that delivers very fast restoral times (under 50 milliseconds) in the event of a fiber cut or other outage. It also supports very low levels of latency and jitter, because end-to-end circuit-switched traffic is not processed at intermediary nodes. But it’s expensive: bandwidth is nailed up on a circuit-by-circuit basis (reliable, but capacity-inefficient).

Ethernet has so far been missing the ability to provide business-class reliability and to manage latency and jitter QoS metrics for voice traffic.

So the Ethernet community has developed IEEE 802.17, or RPR, to blend the characteristics of the two technologies. The RPR mantra is “SONET performance at Ethernet economics!”

Note that prestandard RPR implementations from different vendors – Cisco, Nortel, and others – have different brand names. RPR defines a new media-access control (MAC) packet format that enables statistical multiplexing over a SONET infrastructure (also called “spatial reuse”). The customer side of CPE could “speak” Ethernet, while the MAN/WAN side could “speak” RPR, explains Martin Green, product manager of high-end routing at Cisco.

To handle latency and jitter, RPR allows for two classes of service: high priority and low priority, Green explains. “It’s a simpler way of doing QoS, because, like SONET, it is end-to-end, with no processing in intermediary nodes.”

Note, though, that if there is a fiber cut, while restoral times are on a par with SONET’s sub-50-millisecond failover, there is temporarily less bandwidth available. This is unlike SONET, which leaves 50% of its bandwidth idle for just such situations.