AlcaLu: PBT Is Peripheral

Provider Backbone Transport (PBT), the controversial new flavor of Ethernet that's attracting a lot of industry interest, is a peripheral technology that's not important enough to merit serious attention, according to Alcatel-Lucent (NYSE: ALU). (See PBT Stars at Ethernet Expo , PBT: New Kid on the Metro Block, PBT Gathers Support, and PBT's Ethernet Appeal.)

Lindsay Newell, director of marketing for Alcatel-Lucent's IP division, says his company is seeing little evidence of real interest in the technology. "Our overall position in neutral… If we see widespread carrier demand, we will take a look at it, but we're seeing curiosity at the moment, not demand. We are taking an interest in how this will develop, but our focus is on Ethernet over MPLS, where there is tremendous demand," he says.

"MPLS is the technology that enables carrier Ethernet today," adds Newell, and whether PBT can develop into a carrier-grade technology "in the future is an interesting question, one that will be answered in the next two to three years. PBT is a potentially interesting technology, but IP/MPLS is what is being deployed today and is at the center of our discussions with carriers today."

That "neutral with a negative twist" stance won't come as much of a surprise to many. AlcaLu needs to remain neutral so as not to antagonize BT Group plc (NYSE: BT; London: BTA), a major customer and PBT's main carrier supporter at present. (See BT Pressures Vendors Over PBT, Nortel on PBT: Today BT, Tomorrow the World!, Nortel, Siemens Win PBT Deals at BT, BT Rethinks 21CN Core Strategy, and BT Likes Nortel's New Ethernet Flavor.)

At the same time it doesn't want to encourage interest or uptake of PBT, because it wants carriers to invest in IP/MPLS infrastructure instead, especially as it has gained significant momentum in that market and is investing in new products and upgrades. (See AlcaLu Boasts Router Success.)

Unsurprisingly, the other main metro IP router vendors, Cisco Systems Inc. (Nasdaq: CSCO) and Juniper Networks Inc. (NYSE: JNPR) -- also both suppliers into BT's 21CN next generation network -- are adopting a similar position.

Cisco is tracking PBT's development, but is keen in briefings to highlight PBT's shortcomings when compared directly with MPLS. (See Cisco Tracks PBT Standards Process and Huawei Joins PBT Fan Club.)

AlcaLu is also keen to highlight any potential pitfalls, and takes some swings at one of PBT's alleged benefits, that of management and provisioning simplicity. Newell says "there's a history of technologies that look very simple becoming very complex" as they work their way through the standards process, which PBT is doing at the Institute of Electrical and Electronics Engineers Inc. (IEEE) under the name PBB-TE (Provider Backbone Bridge Traffic Engineering). "There are already signs of feature creep, such as the use of GMPLS as a control plane technology," notes Newell.

Juniper, meanwhile, takes a less aggressive stance. In an email to Light Reading, the company says: "Still in its infancy (and in the early stages of the standardization process), PBT currently provides a small subset of MPLS functionality that has evolved over the years to solve real service provider requirements. Juniper Networks has been monitoring PBT developments and will continue to do so as the technology matures and standards emerge. We remain true to our strategy of delivering to our customers the best, most cost-effective suite of NGN infrastructure solutions."

Like Cisco, AlcaLu prefers to talk about MPLS rather than PBT. When asked about the impact of BT's plans to deploy PBT in 21CN, Newell notes: "BT's decision has attracted a lot of attention, but in Europe alone there are a lot of Ethernet service deployments based on IP/MPLS -- Orange (NYSE: FTE), Telefónica SA (NYSE: TEF), Belgacom SA (Euronext: BELG), KPN Telecom NV (NYSE: KPN), Telia Company -- and we continue to see further widespread deployments based on IP/MPLS."

Newell even says one of the first carrier announcements made by PBT's leading vendor, Nortel Networks Ltd. , involves a peripheral deployment alongside what is a much larger MPLS rollout. He points to the announcement that PBT is set to be deployed by China Telecom Corp. Ltd. (NYSE: CHA) subsidiary Shanghai Telecom, an announcement made in October 2006. (See Nortel's PBT Debuts in China.)

Newell says AlcaLu is helping Shanghai Telecom build its metro Ethernet core with 7750 routers and 7450 switches for Ethernet aggregation to service a city zone of about 9 million people, and that the deployment of PBT and its predecessor, PBB (Provider Backbone Bridging), will be in rural, less dense areas that will then feed Ethernet traffic into the MPLS core. (See AlcaLu Wins Shanghai Deal.)

Indeed, Shanghai Telecom chairman Zhang Weihua is quoted as saying: "IP/MPLS is the strategic direction for network transformation and will help us meet the high expectations of our customers."

Nortel, which is busy making the most of PBT's current high profile, takes a less confrontational approach to the situation in China. (See Nortel Says 40-Gbit PBT Coming Soon and Nortel Demos PBT Integration.)

In a statement emailed to Light Reading, Paul Indoo, a member of Nortel's Carrier Ethernet product marketing team, notes: "Nortel can't comment on the specifics of Shanghai Telecom's Carrier Ethernet network deployment, but we'd point out that Shanghai has a population of more than 17 million so it's not surprising that multiple vendors are involved in this network deployment, as is the case for many large networks."

Indoo adds: "Nortel's position in relation to Carrier Ethernet and MPLS has always been that these technologies are complementary… Where service providers have already deployed a MPLS, we expect to see many cases of Carrier Ethernet and MPLS networks interoperating with each other."

— Ray Le Maistre, International News Editor, Light Reading

twelveoaks 12/5/2012 | 3:07:50 PM
re: AlcaLu: PBT Is Peripheral Just recently, thousands of routers went down in NTT. If you are just a web user, you probably can tolerate the outrage. But imagine a financial customer who reserves a fat pipe out of this network. I wonder how can a carrier afford to take such a risk to sign a contract to provide "reliable" service to such customers and promise the 50ms protection switch over as in SONET network. The whole network is down. The only alternative path is to pray and wait.

On the other hand, in SONET, each circuit is statically configured from management plane for cross connection. A circuit is well isolated from others. A failure at one point does not bring the whole network down.

PBT is another form of SONET with some lower cost characters. In the core, it is a statically configured cross connected system.

The dynamics brought by IP routing and signalling protocols are not good for a carrier network at the transport layer. It is a good choice at the service layer such as to set up VoIP calls, and IMS stuff.

At the transport layer, a customer usually asks for a pipe through a network with bandwidth and other QoS requirements. And most importantly, a protection switching capability such as in 50ms. Those customers sign up the service for a long term. Why does the carrier have to deploy costly complex tools such as IP/MPLS to set up a long term static pipe and maintain the soft states of such a pipe by the routing and signalling protocols through out the life of the service? An error in these dynamic protocols can adversely bring down the service, or maybe the whole network as it already happened.
OldPOTS 12/5/2012 | 3:07:49 PM
re: AlcaLu: PBT Is Peripheral twelveoaks,

As you describe PBT, it is slightly less than using ATM with PNNI; manualy setting up of physical Paths (over SONET or PVPs over other formats) and using SPVCS for providing a service circuit. AlcaLU can already do that, so please explain how PBT is different.

paolo.franzoi 12/5/2012 | 3:07:47 PM
re: AlcaLu: PBT Is Peripheral
I found the Opex argument humorous. The problem with the SONET network is the crossconnects. Timing synchronization is not the problem.

twelveoaks 12/5/2012 | 3:07:47 PM
re: AlcaLu: PBT Is Peripheral OldPOTS,

I think the difference lies in the cost. Both the equipment cost and operational cost. PBT combines the low cost Ethernet hardware and the advantages of cross connect. Actually I believe PBT can replace SONET in the long run because the services and qualities it can provide, yet at a low cost. ATM loses its ground not because of its QoS, but because of its complexity, plus the trend to converge to IP at the application layer.

As the network and application evolves, a cross connection based packet network should be favored as a transport network. The problem with SONET, I think, is with its high cost and inflexibility to expand. These problems stem from the synchronization and circuit switching of SONET at the physical layer. As application layer evolves to IP, the synchronization of SONET is not needed (since the application does not care that).

Ethernet cross connect is much more flexible.
t.bogataj 12/5/2012 | 3:07:36 PM
re: AlcaLu: PBT Is Peripheral twelveoaks,

consider a few points:
* PBT does not scale;
* PBT requires complex TE (say, G-MPLS);
* PBT has no inherent QoS capabilities (like bandwidth guarantees), other than those of .1Q prioritization;
* PBT does not provide 50ms failover w/o CFM OAM;
* PBT requires changes to .1ag CFM (unicast CCMs) - not very likely since it undermines the scalability of .1ag CFM;
* PBT does not support P2MP (multicasting);
* PBT CAPEX is the same as PBB CAPEX;
* PBT OPEX is higher than PBB OPEX due to complex TE;
* etc.

Now, where's the advantage of PBT - except that it promises to feel like SDH/SONET?

Good deal for Nortel and BT. Excellent marketing of a PBT "idea". But no more than that.
Sign In