Two new groups are trying to adapt local fiber rings to handle data more efficiently than tried-and-true SONET, and bring down their cost for service providers and companies.

If successful, the groups will put forth a standard within two years that will spell out how the dual fiber ring architecture of SONET can be used without leaving one ring idle as a backup the way SONET does.

This new technology, known as Resilient Packet Ring (RPR), takes advantage of the fact that a lot of data traffic is less time-dependent than voice.

The newly formed Resilient Packet Ring Alliance met recently in conjunction with a committee of the IEEE, with the goal of writing a single RPR protocol that will unite the ongoing efforts of several hardware vendors, including Cisco Systems Inc. and Nortel Networks Corp. These companies and others have technology that falls under the RPR umbrella, but they are proprietary.

The prospective standard will let companies and service providers deploy metropolitan-area optical bandwidth so it can be distributed efficiently for IP, with a focus on Ethernet. Companies can buy less bandwidth because they will waste less. Similarly, service providers will be able to charge less because their networks will be more cost effective. Carriers also will be able to offer differing levels of service priced from premium to bargain basement.

SONET networks are arranged in dual fiber-optic rings with the active channels flowing in one direction only. If the primary ring breaks, the secondary ring is used and traffic flows in the opposite direction.

This nearly instantaneous switchover is one of the main attractions of SONET because it could preserve voice connections. But because much metropolitan-area network traffic is now data, not voice, uniformly low network latency is less critical. Some protocols and applications such as FTP and data backup can withstand delay. For them, SONET supplies more responsiveness than necessary in some cases.

RPR uses both rings all the time. If one ring breaks, all traffic goes on the other ring. So depending on traffic load, the remaining ring may get congested. To deal with this logjam, RPR introduces quality-of-service parameters so top priority traffic gets the bandwidth it needs and is unaffected by a break, much as it is in SONET. Lower priority traffic, however, may suffer delays.

Service providers using RPR could charge more for a premium service that survives a break without delays and charge less for a service that results in delays in the event one ring fails.

Cisco has pushed RPR gear it calls Dynamic Packet Transport gear for nearly two years, and says it has customers in China. Swedish vendor Dynarc has sold gear in that country using its version, Dynamic Synchronous Transfer Mode, for about the same period. Upstart Lantern Communications plans to trial its Resilient Optical Packet Ring gear with service provider Global Crossing later this year.

Nortel says its Optical Ethernet gear falls under the category of RPR and enables faster provisioning of metropolitan services including letting companies modify their bandwidth as they need more or less.

» posted by abennett

Network World

• Email
• Print
• Share
  • Slashdot
  • Digg
  • Stumble
  • Reddit
  • Newsvine