Ericsson Retraces Packet-Optical Steps
That means the SPT 2700, Ericsson's 2008 attempt at P-OTS, will be discontinued, says Matthew Smith, head of Ericsson's optical product marketing.
"Since we launched that product, a lot has changed," Smith says. "Sometimes, you need to make the call, when one product line isn't going the way you want it to."
One distinction is that the SPT 2700, announced late in 2008, was built for metro core requirements and wouldn't scale down to the shallower metro regions, Smith says. In the intervening year, it's become apparent that carriers want to have options smaller than Ericsson's 32-port box. (See Ericsson Makes Packet-Optical Play.)
Another factor was that the SPT 2700 was built on the assumption that reconfigurable optical add-drop multiplexers (ROADMs) would be a big part of packet-optical convergence. And so far, they just aren't, says Sterling Perrin, an analyst with Heavy Reading
"Ericsson, I'm guessing, was more of a true-to-form God Box, which would have been OK if ROADMs were essential. But in the past six months, it seems the ROADM part isn't as important," Perrin says.
The market in general seems to want a more modular approach, where entire functions can be omitted but brought into the system later. That's certainly true of ROADMs, which Perrin still considers a required option in metro P-OTS.
"A lot of the European companies aren't flocking immediately to the packet-optical converged transport hype," says Eve Griliches, an analyst with ACG Research. "They're still dealing with separated networks and certain products doing certain things."
It doesn't appear the 2700 ever shipped for revenues. The box was in trials, but it's not clear how far along those trials were, Perrin says.
The change of heart points Ericsson in the right P-OTS direction but sets the company back, as the 2700 was supposed to ship this year.
In with the new
Ericsson's new P-OTS offering, announced today, is the OMS 1460. Availability dates remain vague; Ericsson says the box will emerge sometime this year, but that probably drops Ericsson out of contention for the Verizon Communications Inc. (NYSE: VZ) packet-optical request for proposals (RFPs), Griliches says. (See Ericsson Launches Packet-Optical Box.)
But the announcement gives Ericsson a more definitive voice against the swarming P-OTS competition, which -- if you combine metro and core P-OTS offerings -- includes Alcatel-Lucent (NYSE: ALU), Ciena Corp. (NYSE: CIEN), Cyan Optics Inc. , Fujitsu Network Communications Inc. , and Huawei Technologies Co. Ltd. . (See AlcaLu Joins the War for the Optical Core, Ciena Catches Packet/Optical Convergence Bug, Cyan Plays God With Optical, and Huawei Intros Big Crossconnect .)
The OMS 1460 is targeted at metro aggregation and grooming. It follows the roughly one-year-old OMS 1410 Ethernet switch and positions the 1400 family as Ericsson's metro arsenal of the future.
The system segregates TDM and packet traffic when it comes to switching. That's the tactic taken by Ciena with the hybrid fabric in its 5400, while Alcatel-Lucent touts its 1870 TransportTera Switch -- more a core box than a metro one, admittedly -- as having a unified fabric.
The OMS 1460's packet switches support cell-based switching, too, meaning they can also handle Optical Transport Network (OTN) traffic, Smith says. In fact, OTN is going to be a key element of Ericsson's packet-optical plans -- and particularly crucial to the core P-OTS system the company is planning.
"We will have a multi-terabit core switch -- that's clear -- and it will be OTN-based. We've been a fan of the OTN standard for a long time," Smith says.
OTN is one of the requirements Heavy Reading lists for core P-OTS. (See Redefining P-OTS.)
Growth by degrees
Ethernet cards for the OMS 1460 sport their own 20Gbit/s switch fabrics. The distributed approach saves power by avoiding the need for one large Ethernet switch somewhere in the system, Smith says. The little switches connect to each other through a matrix that sits alongside the system's TDM switch.
Similarly, the optical capabilities are meant to be added in trickles, as most operators don't expect to implement them until later, Smith says. In particular, this plays into the trend of not needing ROADMs right away.
"People only require two to three ways of ROADM in the early days, and the other products may be overdimensioned when they put in support for nine ways," Smith says.
— Craig Matsumoto, West Coast Editor, Light Reading