x
Ethernet equipment

Vendors Clash Over PBT

PBT (Provider Backbone Transport) has solidified its reputation as the telecom industry's most controversial technology, with its supporters and detractors jostling for the chance to talk up or shoot down its potential. (See PBT: New Kid on the Metro Block.)

With most of the carrier community watching from the sidelines, equipment vendors have argued vociferously about the validity of PBT (the pre-standards connection-oriented flavor of Ethernet) as a technology worthy of commercial deployment, with the pro- and anti-PBT camps clashing again at the Carrier Ethernet World Congress in Geneva last week. (See PBT vs MPLS: Round VII.)

There, Nortel Networks Ltd. , Ciena Corp. (NYSE: CIEN), Extreme Networks Inc. (Nasdaq: EXTR), Meriton Networks Inc. , and Nokia Networks were PBT's most vocal supporters, while the leading MPLS vendors -- Alcatel-Lucent (NYSE: ALU), Cisco Systems Inc. (Nasdaq: CSCO), and Juniper Networks Inc. (NYSE: JNPR) –- were the technology's major detractors in presentations and panel sessions. (See Nortel Touts Ecosystem and Extreme Tests PBT.)

That division exists because PBT is regarded as an alternative to MPLS in metro networks, a tension highlighted in the latest Light Reading Insider, MPLS vs. PBT: The Empire Strikes Back.

In that report, author Simon Sherrington finds the MPLS vendors under pressure from advances in Carrier Ethernet technology, and having to fight hard to "justify the role of their technology in metro and access infrastructures." (See MPLS: Metro a No-Go?.)

The end result, he feels, is that the emergence of PBT as an option for carriers will drive MPLS innovation and new product development -- something that will ultimately have a positive impact on the router vendors.

Validating multivendor deployments
Pressure on the MPLS vendors intensified in Geneva, where German test facility European Advanced Networking Test Center AG (EANTC) replicated a metro and access network using technology from 24 different vendors. (See Hammerhead Demos PBT, WWP Combines MPLS, PBT, RAD Plays Ethernet Test Role, Gridpoint Plans PBT, Soapstone Goes to CEWC, and TPack Talks Ethernet.)

The demonstration network showed how Ethernet services could run end-to-end across network clouds that were using three different transport and aggregation technologies –- MPLS, PBT, and Transport MPLS (T-MPLS), a stripped down version of MPLS that is similar in nature to PBT. (See EANTC Preps MEF Demo.)

The resulting white paper is available at EANTC's Website..

The challenge set by constructing that test network, notes EANTC managing director Carsten Rossenhoevel, was "to interconnect the three metro clouds to an MPLS core" and run services across the clouds. "Interworking between MPLS and PBT was more or less seamless," says Rossenhoevel.

Having conducted numerous network tests involving many fixed and wireless technologies previously, the EANTC man was surprised by the success attained with a non-standard technology. "Successful interoperability with PBT at such an early stage is very reassuring," he says. "This has been a milestone" for the technology.

Independent evidence that PBT and MPLS can co-exist will be reassuring for carriers looking to deploy the new flavor of Ethernet, and none more so than BT Group plc (NYSE: BT; London: BTA), the one major carrier to have thrown its weight behind PBT.

BT has worked with Nortel on developing PBT, is involved in the standards development work at the Institute of Electrical and Electronics Engineers Inc. (IEEE) , has already deployed PBT in Italy, and is working towards deployment in the U.K. (See BT Goes Live With PBT.)

PBT survey reveals carrier indecision
But while the pro-PBT camp has gathered increasing support, and other major carriers are known to be checking out PBT's potential, the technology is still regarded as a risk by many carriers, while vendors with PBT solutions admit that the current interest shown by operators is largely based on curiosity. (See Deutsche Telekom Flirts With PBT, Nortel Lands More PBT Action, and Verizon Preps God Box RFP.)

A survey of carriers conducted by Synergy Research Group Inc. shows that a vast majority, around 70 percent, are "uncertain" about whether they might deploy PBT, while less than 25 percent stated they were definitely interested in deploying the technology. A small percentage, less than 10 percent, had no interest in PBT at all.

Synergy's research showed that the lack of a standard was PBT's major negative. The technology is currently going through the standards process at the Institute of Electrical and Electronics Engineers Inc. (IEEE) , where it is known as PBB-TE (Provider Backbone Bridging - Traffic Engineering), but is not expected to be ratified until some time in late 2008 or during 2009.

In the meantime, PBT continues to attract further vendor support. (See Ciena Falls in PBT Camp, Bay Adds PBT, and Lightstorm Charges Into Ethernet.)

— Ray Le Maistre, International News Editor, Light Reading

<<   <   Page 2 / 3   >   >>
mvissers 12/5/2012 | 3:01:18 PM
re: Vendors Clash Over PBT PBB requires a lot of configuration to get any service up and running. PBB is the first ethernet network with two layers: backbone service and B-VLAN (i.e. a tunnel layer).

Each PBB network domain contains a few B-VLANs which you need to set up on top of a MSTP environment (which requires configuration itself). Each B-VLAN can have hundred or more end points, which need to be configured before MVRP can register the B-VLAN within the PBB network domain.

Once the B-VLANs are up, individual service instances can be added. This requires manual configuration per service instance of S-VID to I-SID relationships in the I-BEB nodes and manual configuration of I-SID to B-VID relationships in the B-BEB nodes. Inter-domain services may also require the configuration of I-SID Translation in B-BEB nodes. Without this configuration there is no service.

To limit flooding in the B-VLAN, MMRP must register either for every multipoint service instance a service instance specific group address, or the provider must configure a specific group address shared by multiple service instances. The latter requires that MAC Address Translation is configured in B-BEB nodes for such service instance.

PBB backbone service layer does not have a control plane and requires manual configuration of every service instance in the I-BEB and B-BEB nodes.
macster 12/5/2012 | 3:01:20 PM
re: Vendors Clash Over PBT macster: "Why the PBB (802.1ah) comparison with PBT (802.1aq)? Both PBB and PBT will be able to co-exist on the same kit - we can set the VLAN ranges for PBT."

Exactly! So you agree that hardware/fabric for both PBB and PBT is practically the same. Hence, no difference in price. Do you need any control-plane equipment for PBB? None. Do you need a costly control-plane equipment for PBT? -Yesss. Sum that up.

tata, T.

=====================

Sum this up!

PBB is not conection-oriented. PBT is. T-MPLS is. Both will need a control plane - this is where the comparison needs to be done!!!
t.bogataj 12/5/2012 | 3:01:20 PM
re: Vendors Clash Over PBT dask: "...maybe you don't need special control-plane equipment for PBB but you do need control-plane such as STP for it..."

Agree. Yet the main (functional) difference between PBB and PBT is the absence of control plane mechanisms in the latter. No active learning; no active topology protocols like xSTP... not to mention multicasting support.

In PBB transport equipment, a control plane is already built-in.

Clearly, provisioning is the key for PBT. But it is offloaded to an abstract and costly piece of external hardware.

T.
dask 12/5/2012 | 3:01:23 PM
re: Vendors Clash Over PBT > Do you need any control-plane equipment for PBB? None

Hmm, ...maybe you don't need special control-plane equipment for PBB but you do need control-plane such as STP for it, much like you need it for PB/QiQ.
CoreRouterBuilder 12/5/2012 | 3:01:24 PM
re: Vendors Clash Over PBT Please show me your calculation for 30%. I'd like to see it. Is it a public domain info or your own estimate?

Could you put a URL here for a doc describing your 30% cost estimate?
t.bogataj 12/5/2012 | 3:01:24 PM
re: Vendors Clash Over PBT macster: "Why the PBB (802.1ah) comparison with PBT (802.1aq)? Both PBB and PBT will be able to co-exist on the same kit - we can set the VLAN ranges for PBT."

Exactly! So you agree that hardware/fabric for both PBB and PBT is practically the same. Hence, no difference in price. Do you need any control-plane equipment for PBB? None. Do you need a costly control-plane equipment for PBT? -Yesss. Sum that up.

tata, T.
macster 12/5/2012 | 3:01:24 PM
re: Vendors Clash Over PBT Firstly, you really need to understand what PBT and T-MPLS are? When do we use PBT? When is T-MPLS the better option?

The main point is composition: services composition, network composition, etc. Are we only going to have Ethernet services? Or a mix of legacy ones? Do we already have an MPLS core? Are we a mobile operator whi just needs Ethernet connectivity for our RAN and core
(and not M-PBN)?

BT's use of PBT can be seen as being V2 of their 21CN initiative. Why have EoSDH or Dry Martini EoMoS)? Why not Ethernet end-to-end (drawing WDM as far out as possible)? Ethernet UNI, EFM (802.3ah) and Ethernet tunnels to the MPLS core?

Kit-wise, it is the considered wisdom that PBT gear is cheaper than MPLS gear and T-MPLS gear???). I believe 30% is a figure widely mentioned. But it's difficult to give an estimate on OPEX until we have a sizeable PBT and T-MPLS network in operation to compare.
macster 12/5/2012 | 3:01:24 PM
re: Vendors Clash Over PBT Why the PBB (802.1ah) comparison with PBT (802.1aq)?

Both PBB and PBT will be able to co-exist on the same kit - we can set the VLAN ranges for PBT.

???
ROBERT-DELANGE 12/5/2012 | 3:01:26 PM
re: Vendors Clash Over PBT Looking this over closely, it seems PBT just shifts the cost from CapEx to OpEx with a loss of functionality, flexibility and security. I would expect CapEx and OpEx on MPLS-based networks decline as a natural part of its evolution, while CapEx of PBB increases as needed functionality and control is added - first T-MPLS then GMPLS. What am I missing?
t.bogataj 12/5/2012 | 3:01:26 PM
re: Vendors Clash Over PBT Comparing PBB and PBT, there can be no real grounded difference in price: they both require the same HW, switching fabric, etc.

But!

The cost of PBT transport equipment may remain comparable to PBB, but there are hidden costs. The very thing that is willingly overlooked by PBT prophets is the cost of PBT's control/management plane, which has an enormous impact on Capex for PBT deployment. So far, GMPLS is the main candidate for PBT's control plane. Does it come cheap?
<<   <   Page 2 / 3   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE