x
Optical/IP

Optical Core Challengers: Rise Up!

The rise of packet-optical transport systems (P-OTS) is more than just an excuse to create buzzwords and acronyms -- it's also a rare chance for competitive optical equipment vendors to topple the incumbents.

In the core optical space, the top marketshare spots are well cemented by Alcatel-Lucent (NYSE: ALU), Ciena Corp. (NYSE: CIEN), and Huawei Technologies Co. Ltd. , with a combined 82 percent share in optical crossconnects. But that could change as the core P-OTS market grows, according to Heavy Reading analyst Sterling Perrin.

That's because P-OTS represents a rethinking of the optical core. Carriers want the packet and optical networks to converge, and simply upgrading a crossconnect isn't going to do it.

"The onus is on the incumbents to produce next-generation products that are best in class. Otherwise, operators are prepared to bring in new suppliers to meet their core packet-optical transport needs," Perrin writes in a recent report, "The Core Packet-Optical Transport Evolution."

That's why there's a bona fide optical startup, Cyan Optics Inc. , in the P-OTS space. (See Cyan Plays God With Optical.)

Of course, the incumbents aren't taking this lying down. Alcatel-Lucent made its core P-OTS bid Thursday by announcing the 1870 Transport Tera Switch, due to ship in the second quarter of this year. Ciena had previously announced its next-generation optical system, the 5400 family. (See AlcaLu Joins the War for the Optical Core and Ciena Catches Packet/Optical Convergence Bug.) AlcaLu's 1870 didn't get released in time to be included in Perrin's report, by the way. Other core P-OTS vendors discussed in the report include Huawei, Hitachi Ltd. (NYSE: HIT; Paris: PHA), Infinera Corp. (Nasdaq: INFN), and Tellabs Inc. (Nasdaq: TLAB; Frankfurt: BTLA).

Crossconnect vendors missing from that list so far include Ericsson AB (Nasdaq: ERIC) and Sycamore Networks Inc. (Nasdaq: SCMR).

Ericsson already plays in the P-OTS segment for metro networks, with the SPT 2700, but it's yet to develop a core product. The distinction matters, because core P-OTS requires a larger switching capacity -- at least 1 Tbit/s, by Perrin's definition. (See Ericsson Makes Packet-Optical Play and Redefining P-OTS.)

Most optical crossconnect vendors will probably play the core P-OTS game, because the new systems have to be substantially different from what's out there today. Those that don't, however, "will be quickly left behind," Perrin warns.

— Craig Matsumoto, West Coast Editor, Light Reading

HSalemi 12/5/2012 | 4:44:21 PM
re: Optical Core Challengers: Rise Up! Given that the projection of traffic in networks is going higher and higher. Why not manage the core at 10G/ODU2 rather than 1G/ODU0. Is there any specific problem out there that requires switching at 1G rather than 10G?
Pete Baldwin 12/5/2012 | 4:44:20 PM
re: Optical Core Challengers: Rise Up!

You're right in most cases, but there's a common belief that the core will have to switch GigEs too, as Sterling Perrin notes in his report.  A lot of core traffic *will* be managed at 10G, as I understand it -- but carriers want a GigE option too.


I don't 100% know the root cause of that need, but I'd guess it has to do with the amount of video that's expected to travel in GigEs.

obaut 12/5/2012 | 4:44:19 PM
re: Optical Core Challengers: Rise Up!

OK, these packet-optical cross-connects support L0 WDM, L1 TDM (at 1Gbps and/or 10Gbps signal rates) and L2/L2.5 packet level switching. Sounds like mechanical integration of good old technologies that could have been integrated in same products for at least a decade or longer. So what’s the big deal?  (The differences between L1 switching granularity levels seems somewhat trivial in the bigger picture of architectural directions.)

Compared to the model where ASP, CDN or enterprise would operate pure-play L3/2 routers/switches and use leased transport service that a network provider delivers with L1/0 muxes (using a separate peer/carrier L3/2 routing/switching systems on top of the L1/0 transport only where needed), the packet-optical model offers a monolithic L2-0 switching alternative. The network operator now has a choice between two relatively simpler layers per the traditional architecture, and the integrated but more complex packet-optical fabric. Which is better: two simpler layers, or one more complex integrated fabric?

Guess the answer depends heavily on whether integration of the packet-optical technologies delivers more than just mechanical product and management system integration -- Is there something new at optical, digital or packet layers or in their interplay that now brings new architectural efficiencies?

Sterling Perrin 12/5/2012 | 4:44:19 PM
re: Optical Core Challengers: Rise Up!

I know that Verizon was originally thinking of ODU2 as the base switching rate for the core but later amended it to ODU0 (with no interest in ODU1 at all by the way). Other operators that are interested in OTN switching are rallying around the ODU0 rate as well.


Core transport rates will be 10G and above but operators are interested in ODU0 within the optical switch, I think mainly for efficient traffic grooming. A lot of video traffic (and other traffic) will originate on Gig Es, so operators need the ability to dig down to the GigE level to efficiently pack, upack, and route this traffic through the core. Again, ODU0 does not imply that there will be 1 GigE interfaces on these core switches.


Sterling

chechaco 12/5/2012 | 4:44:19 PM
re: Optical Core Challengers: Rise Up!

I recall one research paper that presented well reasoned scenario of metro and core network capasity buildup. According to authors metro networks will experience another 3-5 years of faster than core capasity growth. After that growth of uncachable video traffic (everyone is TV crew and video content producer) will spur buildup in the core. These streams might be aggregated better than 1G but it could be provider's prefer to be better safe than sorry.


And RE: the report and list of P-OTS vendors. I'd be surprised if Fuijutsu Flashware 9500 is missing from it.

HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE