x
Optical/IP

AT&T Boosts Ciena, Cisco

AT&T Corp. (NYSE: T) announced today that it has begun deployment of its nationwide optical network using gear from Ciena Corp. (Nasdaq: CIEN) and Cisco Systems Inc. (Nasdaq: CSCO) (see AT&T Deploys Optical Net).

The announcement is important for a number of reasons. One, it shows that large carriers are still spending money on building next-generation optical networks. This provides some optimistic news in the face of the recent rash of carrier bankruptcies and capital spending cuts (see Global Crossing Falls Overboard, Carrier Pile-Up Claims McLeodUSA , and The Capex Cloud).

Secondly, because AT&T is one of the largest and most sophisticated carriers, the announcement sends strong signals about the types of products and architectures favored in these next-generation networks -- validating the products from Ciena and Cisco.

George Gawrys, AT&T Labs transport network planning director, says that the network is far from complete and that he expects deployments in more than 60 more cities in the next couple of years. The contracts themselves could surpass $100 million for each company over the next few years, according to Rick Schafer, an equities analyst with CIBC World Markets.

Ciena’s CoreDirector is being used for the optical switching portion of the network, says Gawrys. CoreDirector’s ability to groom down to STS1 (51.8 Mbit/s) was one important criterion used to select the product, he adds. Gawrys says the carrier had looked at gear from Ciena competitor Tellium Inc. (Nasdaq: TELM) but decided against further evaluation because it lacked STS1 grooming.

This piece of information seems to confirm a market trend, which is that grooming functionality has become a centerpiece in the marketing wars among optical switching companies (see Tutorial on Grooming Switches and Ciena and Tellium Go Feudin' ).

The Cisco ONS 15454 Sonet Multiservice Platform will be used to provide multiservice functions at the edge of the network such as aggregating lower-rate customer traffic up to high-speed OC48 (2.5 Gbit/s) or OC192 (10 Gbit/s) pipes. AT&T has already deployed more than 100 ONS 15454 systems, and Gawrys says the carrier plans to deploy many more over the next couple of years as the network is built out.

AT&T's choice of the Cisco Sonet product comes as a bit of a blow to Ciena, which has been trying to catch up to Cisco in the next-generation Sonet market. Ciena tried to break into the market with the K2 platform it acquired from Cyras last year (see Ciena To Buy Cyras for $2.6 Billion).

On the long-haul front, AT&T had already announced back in 2000 that NEC Corp. (Nasdaq: NIPNY) was providing the long-haul optical transport piece of the network (see AT&T to Deploy NEC DWDM System).

AT&T's optical network, which has been carrying live traffic since December 2001, already connects more than 40 cities across the United States. Both Ciena and Cisco started deploying their gear back in June 2001.

The financial community downplayed the announcement, saying the information had already been built into the market.

“Everyone knew this was going on for a while,” says Alex Henderson, an analyst with Salomon Smith Barney. “This isn’t really what I’d call a new contract. What’s most surprising is how much they’ve already shipped.”

Investors seem to have thought otherwise. This afternoon Ciena was up 0.84 (8.88%) to 10.30, and Cisco was up 0.76 (0.75%) to 17.51.

CIBC's Schafer says he isn’t changing his Hold rating on Ciena, which warned the Street last week that its earnings would be down this quarter (see More Cuts at Ciena). But Schafer thinks the news is giving the stocks a short-term boost, because the contract is finally out in the open. Glenn Jasper, Ciena’s director of public relations, says this is the largest deployment of CoreDirectos in an incumbent carrier's network to date. Cisco also says that this is one of its largest deployments of the ONS 15454 product. While it claims that it has also shipped to several other large inter-exchange carriers, this is the first large carrier contract it has announced publicly.

— Marguerite Reardon, Senior Editor, Light Reading
http://www.lightreading.com
Page 1 / 3   >   >>
wdog 12/4/2012 | 10:57:38 PM
re: AT&T Boosts Ciena, Cisco "The Cisco ONS 15454 Sonet Multiservice Platform will be used to provide multiservice functions at the edge of the network such as aggregating lower-rate customer traffic up to high-speed OC48 (2.5 Gbit/s) or OC192 (10 Gbit/s) pipes."

I can see where the 15454 fits in aggregating traffic from the metro across the core, but what about traffic from data centers across the core. This would likely be coming off an OC-48 or OC-192 link to a core router, right? Does anyone have ideas on how the Avici router fits in the picture?
let-there-be-light 12/4/2012 | 10:57:35 PM
re: AT&T Boosts Ciena, Cisco While I'm not familiar with the details of the AT&T network, I don't think there is any reason why the IP part of the traffic being aggregated by the Cisco box would not be fed into a core router such as the Avici and from there directly across the WDM core to other core routers.

Perhaps I misunderstood your question.
wdog 12/4/2012 | 10:57:33 PM
re: AT&T Boosts Ciena, Cisco This is why I asked the question. My assumption is that CoreDirector will be a layer on top of the DWDM core providing the ability to provision lower speed circuits over DWDM lambdas. In that case, the Avici routers would be interfaced to the CoreDirector, not directly over lambdas.
copperdog 12/4/2012 | 10:57:32 PM
re: AT&T Boosts Ciena, Cisco It is my understanding that T will only be using the CD's on the voice portion of the network, which runs at OC-48. The CD's will not be used in the data portion of the network because it runs at OC-192. I believe that they are expecting to allow the routers to do the restoration, which may or may not work. Ciena has been planning on the 192 card but is still having some issues providing high speed restoration because their systems groom at STS-1. It appears that one of their advantages, may also be somewhat of an achilles heal. JMO
let-there-be-light 12/4/2012 | 10:57:30 PM
re: AT&T Boosts Ciena, Cisco wdog,

Thanks for the clarification. Actually, the thought of sending a 2.5G or 10G concatenated stream of data into a CoreDirector never occurred to me.

Seemed a bit like shredding papers to ease putting them into boxes, and then trying to re-construct the documents later. I could be wrong, though.

copperdog, when you say "voice", I guess you mean leased lines also (which could be carrying data, right?!)...
wdog 12/4/2012 | 10:57:29 PM
re: AT&T Boosts Ciena, Cisco "Thanks for the clarification. Actually, the thought of sending a 2.5G or 10G concatenated stream of data into a CoreDirector never occurred to me.

Seemed a bit like shredding papers to ease putting them into boxes, and then trying to re-construct the documents later. I could be wrong, though."

I agree if it is an OC-48C or OC-192C concatenated stream it doesn't make a lot of sense. That's the reason for the question, If T is spending the money to build this transport network with STS-1 grandularity, I would expect they would want to be able to manange data as well as voice traffic at that level, so wouldn't they expect devices feeding traffic into the network (ADMs, routers, maybe ATM switches?), to support lower levels of grandularity as well? Since data represents the higher volume of traffic, and certainly will continue to outpace traditional voice, why not just go with a non-STS-1 grooming solution like Tellium if the main source of data traffic is just going to feed the transport network concatenated streams?
let-there-be-light 12/4/2012 | 10:57:28 PM
re: AT&T Boosts Ciena, Cisco Yep,

I think (but once again, I'm not very familiar with AT&T's network) that AT&T is still making A TON OF MONEY with lower-speed leased lines (which in turn is being driven by growth in data). That is why these CoreDirectors are worth their weight in gold. Perhaps we'll get a response from someone who could confirm this.

As to whether CD's can handle the switching of all speeds, I'll believe you, Buttercup, if you say so, but I doubt whether it's the optimized solution, particularly for concatenated OC-192.
Buttercup 12/4/2012 | 10:57:28 PM
re: AT&T Boosts Ciena, Cisco Note that CoreDirector has STS-1 grooming and can also switch any size concatenated signal. Having STS-1 grooming doesn not necessarily mean you have to incur any extra expense to switch an OC-48c or OC-192c - it means you have the *option* to switch anything from an STS-1 up to an OC-192c as a single circuit. A TELM switch is too limiting for an all-purpose core, since the supported granularity is too coarse. The equipment feeding the core has a mix of granularity requirements, and certainly is not all OC-48c/OC-192c.
backnaction 12/4/2012 | 10:57:27 PM
re: AT&T Boosts Ciena, Cisco Gawrys says the carrier had looked at gear from Ciena competitor Tellium Inc. (Nasdaq: TELM - message board) but decided against further evaluation because it lacked STS1 grooming.

(This doesn't help their price any....)
Two 12/4/2012 | 10:57:26 PM
re: AT&T Boosts Ciena, Cisco As to whether CD's can handle the switching of all speeds, I'll believe you, Buttercup, if you say so, but I doubt whether it's the optimized solution, particularly for concatenated OC-192.

Let someone from inside the industry give you a peek at the way things really work. No-one makes money from OC-48 pipes today by selling them directly to consumers. Carriers like Broadwing sell them to other, smaller carriers who use their own equipment to subdivide these pipes down into sell-able units (used mostly for VPN and backhaul traffic which are still DS-3's). If you deploy an STS-1 switch in the core, your customer doesn't have to decide between one or two OC-48's...he can get an effective OC-72 between two points if he wants. You can sell the rest to someone else, or reserve it for protection bandwidth. This saves mucho money.

As far as cost: CIENA has always advertised the fact that they make their own ASICs. These custom chips are the framers and switches that make up the dataplane of their device. This costs a lot of up-front cost (and development risk) to develop, but once that's complete, the incremental cost is less than 1/10th that of their competitor's solution, and is certainly dwarfed by the cost of the optics on the line cards. Hence, from a pure COGS (cost of goods sold) point of view, there is NO APPRECIABLE DIFFERENCE in cost between an STS-1 switch and an OC-48 "core grooming" switch. Don't believe me? Ask Velio how much one of their chips costs...they'll tell you >$3k each. IBM might quote you $200 in volume.

If you think about it, connection management should get easier if the size of each connection gets larger (say OC-192c), so if you're smart, you manage on the connection level rather than the STS-1 level. This way, your switch can appear to be an STS-1 switch if you want it to, or alternatively appear to be an OC-48 "core grooming" (in quotes because it's a made-up term) switch with very little performance penalty.

The CoreDirector team took the risk and made it work. The Tellium team decided to be more conservative and the market is making their choice as to who has the better product.

Let the race begin.

..
Page 1 / 3   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE