x
Optical/IP

Juniper Makes Its Packet-Optical Move

Juniper Networks Inc. (NYSE: JNPR) is joining the packet-optical transport systems (P-OTS) ranks, proposing an architecture that stands out for its emphasis on Multiprotocol Label Switching (MPLS).

Juniper's PTX Series Packet Transport Switch, scheduled to launch Thursday, adopts the philosophy of using MPLS for handling packet traffic and Optical Transport Network (OTN) for circuit traffic. That's in contrast to the OTN focus of other packet-optical products on offer from vendors such as Alcatel-Lucent (NYSE: ALU) and Ciena Corp. (NYSE: CIEN).

"We're saying LSPs [label switched paths] rather than OTN, because we believe LSPs are the future," says Luc Ceuppens, Juniper's senior director of marketing. That packet-based approach is more suited towards the bursty, unpredictable patters of future network traffic, Juniper believes.

The PTX is starting life as an MPLS switch -- what used to be called a Label Switch Router (LSR). As a core box it's a Layer 2 alternative to expensive core-router ports, a concept that Verizon Communications Inc. (NYSE: VZ) has been championing at optical conferences for a couple of years.

Did we mention the PTX is big? Based on a new chipset called Junos Express, the PTX can support 480 Gbit/s per slot. An eight-slot version will be available in July, and a 16-slot version should follow six to 12 months later. Later on, Juniper plans to introduce a multichassis version of the PTX, with arrangements that can handle a claimed 3,800 Tbit/s of total traffic.

Why this matters
As P-OTS has grown in importance, with even Cisco Systems Inc. (Nasdaq: CSCO) announcing a box, the pressure has been on Juniper -- a company with no optical heritage -- to come up with something. And Juniper has decided to make its entrance big, producing a core P-OTS system to compete with AlcaLu's 1870 Transport Tera Switch and Ciena's 5400.

But not right away. Key elements still aren't there: Reconfigurable Optical Add/Drop Multiplexer (ROADM) support will come in 2012 and OTN switching in 2013, says Heavy Reading analyst Sterling Perrin.

"It looks like everybody has the same endgame in mind. They're just coming at it from different strengths in terms of what comes out first," Perrin says. Huawei Technologies Co. Ltd. , for instance, took the opposite tactic, starting with a big optical box that could add packet handling later.

Ceuppens claims the "majority" of Juniper's customers have not asked for OTN. Perrin, on the other hand, points out that practically every telco still has circuit-switched traffic that OTN was designed to support. It's going to be tough marketing to telcos without OTN, he notes.

Cable operators, though, would be more willing to go with an all-packet core. Maybe that's where Juniper's real initial targets for the PTX will have to be.

For more
Here's what's transpired with Juniper and optical networking in the past year or so.

And here's the skinny on core P-OTS.

— Craig Matsumoto, West Coast Editor, Light Reading

Page 1 / 2   >   >>
Sterling Perrin 12/5/2012 | 5:11:18 PM
re: Juniper Makes Its Packet-Optical Move

Juniper didn't directly say that MPLS will compete with OTN. Packet traffic will be handled by MPLS and circuit traffic will be handled by OTN in the PTX. But the architecture does seem to pit MPLS vs. MPLS-TP. Juniper is saying that MPLS is the best way of handling packet traffic - and "TP" is not required.


Given the breakdown in the ITU and IETF work on MPLS-TP, I'm starting to wonder if MPLS-TP could weaken a la PBB-TE. MPLS-TP doesn't seem as inevitable as it did a few months ago. Telcos (like investors) don't like uncertainty. Just some thoughts ..


Sterling

Pete Baldwin 12/5/2012 | 5:11:18 PM
re: Juniper Makes Its Packet-Optical Move

Ah, important clarification. Thanks Sterling.

Bob Saccamano 12/5/2012 | 5:11:18 PM
re: Juniper Makes Its Packet-Optical Move Stratus, Falcon, Space, Pulse, Optical. Along with Trapeze and the other acquisitions. That's a lot for a company the size of Cisco (and we are seeing the results of Cisco's trying to engage in a multitude of projects... not good). We will have to see how Juniper does with having to support so many projects. Given the long build up to Stratus only to then release a simple 10GE TOR at launch may indicate they are biting off way more than they can chew.
Pete Baldwin 12/5/2012 | 5:11:18 PM
re: Juniper Makes Its Packet-Optical Move

This seems a little risky. I'm sure Juniper wouldn't have gone down this direction without consulting customers (as people point out on these message boards, there aren't many of them) -- but it's a bit contrarian, or maybe futuristic, to put MPLS center stage when so many telcos are talking about OTN.


Sterling Perrin had a good point about this: Cable companies, who don't have any Sonet to bother with, might find an all-packet approach attractive. It's the telcos, who still have TDM flitting around their networks, who'll want that OTN support.


Then again, there's also the issue of cutting down on core-router ports. Even without the optical pieces like a ROADM, this could be a useful Layer 2 core option.


By the way -- Juniper does say the PTX will fully support OTN. That is, once the functionality is available, you could use the box as a pure OTN switch and ignore the MPLS parts, if you really wanted to.


[UPDATE: I'd originally included, in this comment, my description of the ADVA/Juniper division of labor on the optical side -- apparently my understanding wasn't correct, so i've axed it.]

Pete Baldwin 12/5/2012 | 5:11:18 PM
re: Juniper Makes Its Packet-Optical Move

Luc Ceuppens, on this product being called "Packet Transport Switch" as opposed to Stratus being called "QFabric:" 


"In telecom, your product name has to say what it does."


He was just joking around, but you know ... he's got a point ...

Pete Baldwin 12/5/2012 | 5:11:17 PM
re: Juniper Makes Its Packet-Optical Move

Yeah, I've been thinking about that too. It's a lot to work with. And it shows why Kevin Johnson was so willing to up the R&D budget in the face of the recession.


I do think there was a lot more in the Stratus launch than just the TOR. I felt like I got a lot out of it, anyway. But, to follow your point: It will be interesting to see how quickly the other pieces ramp up.

Pete Baldwin 12/5/2012 | 5:11:16 PM
re: Juniper Makes Its Packet-Optical Move

OK, clarification from Luc Ceuppens on the OTN side:


The PTX is built for a packet/MPLS world, but Juniper realizes there's more to life than packets. The intent, then, is to keep circuit traffic in circuit form, switched via OTN, while handling packets by using MPLS.  Packets and circuits each "stay in their own worlds," Ceuppens says.


On the optical division of labor: What Junpier wanted to point out was that some of the PTX's optical subsystems are coming from providers other than ADVA.  My mistake for saying in an earlier comment that ADVA did all the optical pieces.

ethertype 12/5/2012 | 5:11:16 PM
re: Juniper Makes Its Packet-Optical Move

Sterling, I think you're missing a fundamental point:  MPLS-TP is really not focused on the core, so the lack of emphasis on MPLS-TP in this core announcement says absolutely nothing about its future.


 


MPLS-TP will be deployed in access and aggregation networks first, and it has plenty of increasing traction there.  Portions of MPLS-TP OAM will find their way into the core eventually, but no one will seriously think about building an entire core network around MPLS-TP with static LSPs and NMS-driven provisioning.


 


The other point where you seem confused (or perhaps intentionally provocative?) is that there is no battle to "pit MPLS vs. MPLS-TP."  The whole point of MPLS-TP (and its fundamental advantage over PBB-TE and other options) is that it is compatible with dynamic MPLS.  You can deploy both in the same network, and use them as appropriate.  Saying "MPLS vs. MPLS-TP" is like saying "TCP vs. UDP".

Sterling Perrin 12/5/2012 | 5:11:16 PM
re: Juniper Makes Its Packet-Optical Move

Ethertype -


<Sterling, I think you're missing a fundamental point:  MPLS-TP is really not focused on the core, so the lack of emphasis on MPLS-TP in this core announcement says absolutely nothing about its future.>


On this point, I disagree. I think alot of operators and suppliers are looking at MPLS-TP for the core. Our surveys have shown this.


<The other point where you seem confused (or perhaps intentionally provocative?) is that there is no battle to "pit MPLS vs. MPLS-TP."  The whole point of MPLS-TP (and its fundamental advantage over PBB-TE and other options) is that it is compatible with dynamic MPLS.  You can deploy both in the same network, and use them as appropriate.  Saying "MPLS vs. MPLS-TP" is like saying "TCP vs. UDP".>


On this point: i put that out there as a thought - i have not drawn a conclusion and could be off-base. It may be a nuance, but if suppliers start saying "you definitely need MPLS but you don't need MPLS-TP," then I see them as competing.


Sterling


 

Pete Baldwin 12/5/2012 | 5:11:15 PM
re: Juniper Makes Its Packet-Optical Move

Now, there's something to think about.


I'm at Juniper's financial analyst conference in San Francisco.  Kevin Johnson (CEO) has talked about how the company sets up small incubation groups to investigate new R&D directions.


Pradeep Sindhu (CTO) is following that up by talking about targeting disruptions in different parts of the network:


* Mega data center - Stratus


* Mega POPs - PTX


* Meta COs - the "universal edge" stuff, I think [didn't quite catch this one]


... And Juniper sees potential in the "access and aggregation" part, too.  "In the next two or three years, if we succeed in our incubations, maybe you'll see things here," he said.


Juniper has no plans to go there now. Interesting that they're thinking about it, though.

Page 1 / 2   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE