Cisco Systems Inc. (Nasdaq: CSCO) has taken a big U-turn on its efforts to influence the direction of the proposed standard for resilient packet ring (RPR) networks. Instead of trying to bulldoze folk into accepting its spacial reuse protocol (SRP), a technology already widely deployed by Cisco, it’s dropped the idea altogether and proposed something called “Gandalf.”
Gandalf will require Cisco and its followers to re-spin SRP silicon and won’t give them an inside track in terms of having an existing installed base, according to a presentation they gave to the 802.17 working group of the Institute of Electrical and Electronics Engineers Inc. (IEEE) in Austin, Texas, last week. The presentation makes a big thing out of saying that “Gandalf is not compatible with SRP” and that it’s “not backwards compatible with any existing implementation.”The paper is posted on the IEEE802.org Website at http://www.ieee802.org/17/documents/presentations/nov2001/sw_over_03.pdf
Contributors to the Gandalf paper include Cisco and other vendors already making SRP products, such as Riverstone Networks Inc. (Nasdaq: RSTN) and Mindspeed Technologies.
Other contributors include Agere Systems (NYSE: AGR), Sun Microsystems Inc. (Nasdaq: SUNW), Infineon Technologies AG (NYSE/Frankfurt: IFX), Fujitsu Network Communications Inc. (FNC), Broadcom Corp. (Nasdaq: BRCM), Redwave Networks, Corrigent Systems, Appian Communications, Lara Networks Inc., and DataNet Associates.
Cisco’s opponents in the IEEE 802.17 working group, which have already joined forces to promote a technology called Aladdin as the basis of the RPR standard, acknowledge Cisco’s efforts. “They’ve shown they're willing to change, and that’s very positive,” says Raj Sharma, director of product marketing at Luminous Networks Inc. and vice chairman of the Resilient Packet Ring Alliance.
Other Aladdin backers include Alcatel SA (NYSE: ALA; Paris: CGEP:PA), NEC Corp. (Nasdaq: NIPNY), Nortel Networks Corp. (NYSE/Toronto: NT), Lantern Communications Inc., Vitesse Semiconductor Corp. (Nasdaq: VTSS), and Scientific-Atlanta Inc. (NYSE:SFA).
Sharma says that Gandalf is a big improvement over SRP because it addresses the fundamental goal of the proposed RPR standard -- to provide a way of enabling carriers to offer three types of service. These are: where bandwidth is dedicated to particular connection; where service-level agreements (SLAs) are offered, but traffic shares bandwidth; and best-effort services. SRP only offers dedicated and best-effort services; it doesn’t incorporate weighted bandwidth reservation, which is required for SLAs.
However, there’s still a fundamental difference between Gandalf and Aladdin, a difference that will be hard to reconcile. Aladdin works on the basis of preventing congestion happening in the first place -- a traditional telecom approach. Gandalf works on the basis of detecting congestion and then dealing with it -- the approach used in router networks. Sharma contends that the latter approach can’t provide the same level of performance guarantees, because there’s bound to be a time lag between detecting and dealing with congestion.
Cisco wasn’t able to field anybody to talk about Gandalf by press time.
It’s worth pointing out that Cisco’s shift in strategy in the RPR standards group probably won’t have a big impact on customers that have already installed its SRP-based equipment. That’s because the technology within rings can be proprietary, so long as the connections to the rings comply with the standard, whenever that finally emerges.
The big question is whether the standard will ever emerge. This latest development hasn’t moved the process forward at all. In fact, it’s probably moved it backward, because Gandalf isn’t much more than an idea right now. For instance, last week’s presentation to the IEEE working group noted that a comparison with Aladdin wasn’t complete because of “a problem with the current simulator.”
All of this may also be a case of fiddling while Rome burns. RPR might be a groovy idea, but a lot of carriers are likely to stick with Sonet to build their metro networks, because it’s proven technology and most of their revenues are generated by traffic travelling across TDM (time-division multiplex) channels. Other carriers are likely to try Ethernet rather than RPR, on the basis of it being backed by a whole industry, driving down costs, coupled with a huge established base in customer sites (see RPR: RIP?).
— Peter Heywood, Founding Editor, Light Reading
http://www.lightreading.comWant to know more? This very topic is the subject of a session at Lightspeed Europe,Light Reading’s annual conference, on December 4-6, 2001, in London. Details: [Metro Networks: Latest Developments