x
MPLS

Cisco's Core Router Goes Packet-Optical

Cisco is announcing Monday that it's adding an MPLS blade to the CRS-3 core router, taking on the recently announced Juniper PTX.

For both companies, it's a nod to the packet/optical convergence everybody is talking about. Optical networking vendors have focused on building packet-optical transport systems (P-OTS), but Cisco and Juniper are proposing a different direction, where packet-based traffic is transported through the core via a really big MPLS switch.

It's worth noting that Cisco does have a P-OTS in the Carrier Packet Transport System (CPT). But it seems likely Cisco will put more emphasis on the MPLS alternative, given that it's part of a flagship core router.

Why this matters
This helps clear up last year's questions about how these two router vendors would respond to the growing P-OTS market. While still supporting the Optical Transport Network (OTN) in spirit, their focus will be on this core MPLS switch, or label-switched router, or packet transport switch, or -- whatever you want to call it.

But Cisco is pulling as much as possible into one box, whereas Juniper's PTX is a separate system that's meant to sit alongside a T-series core router and an optical transport shelf (Juniper offers its own, a rebranded ADVA Optical Networking FSP 3000).

That's going to present some opportunities for one-upmanship. Cisco is already boasting about the benefits of integration. "You don't burn router ports going from a core router to a packet transport box" or an optical shelf, says Suraj Shetty, Cisco's vice president of service provider marketing.

One important factor for both companies is that this label-switched router represents a completely new market for the router vendors and might be the easier direction for packet/optical convergence, says Ray Mota, an analyst with ACG Research .

"I've always viewed that it's easier for the IP guys to get into optical" than the other way around, Mota says. "I see opportunity for both Cisco and Juniper. Optical is like a $12 billion market."

For more
Here's the latest on Cisco and Juniper's packet-optical plans.

— Craig Matsumoto, West Coast Editor, Light Reading

Pete Baldwin 12/5/2012 | 5:08:04 PM
re: Cisco's Core Router Goes Packet-Optical

I'm now curious why Juniper didn't integrate the PTX into its own core routers.


One reason might be scale - maybe they could build a bigger switch by keeping it dedicated?


fwiw, Cisco claims that the CRS-3 can be price-competitive as a pure label-switched router. That is, if you filled it with nothing but these Packet Transport blades, basically making it an MPLS switch rather than a core router, it wouldn't be hideously expensive compared to someone else's label-switched router.


That struck me as odd, but Shetty insists it's true.

photon2 12/5/2012 | 5:08:03 PM
re: Cisco's Core Router Goes Packet-Optical

Well, it makes sense that a chassis fully loaded with the lowest cost boards and the simplest software would be less.  Question is, is that the configuration most folks would use the CRS-3 for?  And, does it have the switching scale and overall capacity the PTX is supposed to have?  Time will only tell.


P2

chechaco 12/5/2012 | 5:08:02 PM
re: Cisco's Core Router Goes Packet-Optical

Craig,


one of the main challenges of building multi-layer P-OTS that integrates packet and TDM (OTN/SONET) is, in my view, switching fabric. Keeping Packet separate from TDM and, possibly, ROADM maintains things so that packet and TDM flows do not intersect over the same switching fabric. Such approach will have certain benefits as well as its own issues when compared to integrated P-OTS solution. In addition to level of integration of transport layers solutions might be differentiated by how transport layers interact among themselves in enabling client service, allocating resources.


Cheers

Sterling Perrin 12/5/2012 | 5:08:01 PM
re: Cisco's Core Router Goes Packet-Optical

Craig,


RE: That struck me as odd, but Shetty insists it's true.


Right. They certainly would not be able to charge more for this configuration vs. a MPLS-only competitor, but I wonder what it does to the CRS-3 margins?


This is exactly the explanation that Juniper gave to me (and financial analysts) for why they launched an MPLS switch as a totally new platform, instead of as cards on the core router line. The router gets high router margins but the new MPLS switch gets lower margins, but addresses a new application. This explanation from Juniper seemed very reasonable to me.


Sterling

Pete Baldwin 12/5/2012 | 5:07:44 PM
re: Cisco's Core Router Goes Packet-Optical

Thanks, Sterling.


btw, I did ask Luc Ceuppens of Juniper about the PTX being separate, and he gave the answer I'd expected: it's a deliberate choice, driven by the need for big scale and a smaller-than-core-router price.


He also said that label-switched router traffic arriving at a core router would still have to go through the core-router fabric, going through more processing than necessary. 


Ceuppens sound bite: "It's like a car dealer telling you he can turn your family car into a race car by giving it bigger tires and a spoiler."

Pete Baldwin 12/5/2012 | 5:07:44 PM
re: Cisco's Core Router Goes Packet-Optical

Photon2 - you're right: probably nobody would buy a CRS-3 and load it up with non-router blades. I'd asked the question partly academically.


Sterling nailed the answer, though - the market won't bear the same price for MPLS as it does for core routing.

obaut 12/5/2012 | 5:07:42 PM
re: Cisco's Core Router Goes Packet-Optical

Are these non-IP-router MPLS LSRs generally inter-operable with legacy IP/MPLS LERs?


How are the MPLS LSPs set up between the L3 LERs across these (L2?) LSRs? Based on MPLS-TP?


Are we, finally, coming to the point were SPs can build L2 MPLS cores and services without the costs and complexities of IP-based LSP management (and their CSCO/JNPR controlled IP-based signaling protocols)?


Consequently, are we now able to realize the main advantages of MPLS-TP, eg per http://en.wikipedia.org/wiki/MPLS-TP: MPLS-TP is expected to be a low cost L2 technology (if the limited profile to be specified is implemented in isolation) that will provide QoS, end-to-end OA&M and protection switching.




HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE