Switch Vendors to Tackle Core Routing

A handful of big data switch vendors are readying announcements of equipment that combines both IP routing and ATM switching in products they claim will help carriers take another step toward revamping their core networks.

Alcatel SA (NYSE: ALA; Paris: CGEP:PA), Lucent Technologies Inc. (NYSE: LU), and Marconi PLC (Nasdaq/London: MONI) have confirmed the imminent release of gear that supports switching and routing in both cell- and packet-based networks. And sources say Nortel Networks Corp. (NYSE/Toronto: NT) is waiting in the wings.

Here are the highlights:

  • Alcatel says the second release of its 7670 Routing Switch Platform (RSP), now in customer trials, is set for the end of next week. New features will include support of a multishelf design, in which the basic switching fabric is isolated from the interface cards. This will enable Alcatel to support 450-Gbit/s capacity with OC48 connectivity (OC192 is due late in 2002), as well as combined ATM switching and IP routing.
  • Lucent says its TMX 880 Multiservice Exchange Switch will ship at the end of January 2002, with OC192 support, ATM switching and IP routing, and 160 Gbit/s of capacity. Lucent's not saying anything about trials just yet, however.
  • Marconi says its BXR-48000 is in field trials with one top-tier carrier, with two more trials expected to start before general availability in March 2002. At that time, the product will support OC48 and feature 240 Gbit/s of capacity, as well as ATM switching. By June, it will have OC192 support and 480 Gbit/s of capacity, as well as IP routing capabilities, the vendor says.
  • Nortel is said to be readying its Passport 20000, the next iteration of its multiservice WAN switch, for first-quarter 2002 release. This box will be equipped with OC192, ATM switching and IP routing, and about 160 Gbit/s of full-duplex bandwidth, sources say. Nortel would not comment.

The trend highlights a new sense of urgency among vendors to lay claim to what they see as a multibillion-dollar opportunity to help carriers preserve incoming revenue from existing ATM and frame relay networks, while moving to IP-based future networks -- ones that will be ready to deploy MPLS Multiprotocol Label Switching (MPLS) to control quality of service (see Lucent Unveils Product Lineup and Alcatel ATM Switch Steps Up).

While the upcoming crop of switches doesn't convert traffic between ATM and MPLS, the vendors hope that supporting both concurrently will at least support carriers interested in having both kinds of networks for the foreseeable future.

But analysts say the jury's out on whether the established ATM switch vendors can pull off their plans. This is primarily because they're taking so much on by aiming to support both switching and routing in one platform.

The vendors say this tack is key. If they support only ATM, they'll be unable to offer the throughput and granularity carriers want in future IP networks, they say. If they support IP only, they'll threaten their existing technology -- which they say carriers don't want to do.

"By supporting both switching and routing in one platform, we give carriers a choice," says Mike Lisanti, director of product management for the BXR at Marconi. "We don't dictate to them one way or the other. We avoid cell and packet tax. We provide full Layer 3 functionality. We don't just tunnel ATM over IP packets. This gives us greater efficiency and scaleability."

Despite these arguments, the strategy's pitfalls are clear. Carriers will need to invest in the new equipment, which will be expensive. There's some argument about the relative merits of MPLS itself (see MPLS Gets Lukewarm Reviews and Has the IETF lost it?). And despite claims to the contrary, there will be a point at which the switch vendors position their wares in direct competition with routers from the likes of Cisco Systems Inc. (Nasdaq: CSCO) and Juniper Networks Inc. (Nasdaq: JNPR).

Sources say one has only to look at the ailing Avici Systems Inc. (Nasdaq: AVCI; Frankfurt: BVC7) to see what's at stake in this strategy (see Avici Warns, Wall Street Scorns). Cisco and Juniper not only have clear dominance in the market, they also have their own core strategies in mind. In addition, Cisco is readying a follow-on to its own Cisco MGX 8850 IP + ATM Multiservice Switch, one that will presumably have capacity equivalent to what the other vendors plan to offer.

"We want to provide core transparency, whether that means packet, cell, or frame," says Don Proctor, VP and general manager of Cisco's Multiservice Switching business unit. Clearly, Cisco's ready to mine its own share of the legacy migration market.

There's also a chance the switch vendors won't be able to pull off the technology fast enough to gain traction. Marconi, for instance, admits that its full roster of core features won't be ready until mid-2002, a full year and half after its first announcement.

The switch vendors also face competition from emerging players like Équipe Communications Corp., which says its Équipe 3200 (É3200) platform allows carriers to move their ATM networks to MPLS functionality without having to install a separate IP infrastructure first (see Équipe: Take the ATM Road to MPLS).

For its part, Équipe, perhaps prudently, is staying out of the routing fray. "We have never said anything about supporting native IP," insists VP of marketing Bob Sullebarger.

Analysts say all this adds up to a major challenge, one the switch vendors will need help to overcome. "The concept of a switch/router is as old as the hills... But if the switch vendors are really serious about routing, they'd better have the inside track with important customers," says Jim Lawrence, program director at Stratecast Partners. "Otherwise, they're asking to become another Ironbridge." (See IronBridge Has Fallen Down, Ironbridge 'Sold for Parts', and Ironbridge's Last Ditch Efforts Fail.)

It's more likely, Lawrence says, that the emerging switches will wind up taking over label switching in MPLS networks, leaving the heavy lifting at Layer 3 to the big router players. This is still a growing opportunity. "Internets, intranets, and extranets are going to have a lot of label-switched paths [to handle]," he says.

Whatever materializes, the established switch vendors seem more than ready to stake their claims, big time, and they're not backing down on the biggest claim of all -- to support IP routing. "We've had a lot of debates internally about what to call [our product]," says Marconi's Lisanti. "ATM switch, router... We will offer carrier-class IP routing." — Mary Jander, Senior Editor, Light Reading
<<   <   Page 2 / 6   >   >>
Belzebutt 12/4/2012 | 7:24:38 PM
re: Switch Vendors to Tackle Core Routing A cynic would say the growth in multiservice has fallen off and this is a deperate leap by all of them to see the world converted to ATM+MPLS which is something those layer 2 products might be able to do, let alone routing.

Even core router sales have fallen this year. There is no gravy train anywhere, but multiservice switches are still selling just like routers. The death of ATM has been proclaimed years ago and yet look at that, it's still going strong. I guess the customer is wrong once again!
flanker 12/4/2012 | 7:24:37 PM
re: Switch Vendors to Tackle Core Routing "Further destabilizing this, or learning how to be a router company ten years after the fact is suicide.

I think the router functionality is an MPLS Trojan Horse, something nobody wants to concede becuase of all the MPLS haters out there. Level3, joining ATT and a lot of other US and European carriers, announced that they have incorporated MPLS into the backbone this week.

rjmcmahon 12/4/2012 | 7:24:36 PM
re: Switch Vendors to Tackle Core Routing I think the router functionality is an MPLS Trojan Horse.

This makes the most sense. Do these MPLS products support path provisioning and QoS such that they can at least meet the basic carrier requirements?

PS. I don't understand why people hate MPLS, though I have never used it so I don't have any real world experience.
Holy Grail 12/4/2012 | 7:24:35 PM
re: Switch Vendors to Tackle Core Routing
Interesting times....

So ATM is going gangbusters? And VoIP is elusive.

IMHO. Native ATM Service never took off, few people have NSAP addresses and even if they did they would be unable to establish an SVC.

ATM was and remains useful as a high speed packet switching transport. Used in the core of the Internet when routers could not support line rate forwarding (not used any more now we have routers with OC192 and line rate forwarding). Still used today in the core of SP carrier networks for transport of Frame Relay Services and Voice Trunking (AAL2).

Is ATM still useful, sure when you consider that most of the SP data revenues come from Frame Relay Services.

So which of these monster ATM switches are likley to be sucessful? Well I guess you need to stop thinking about technology bits and bytes and start looking at who owns the existing Frame/ATM SP infrastructures. In fact it is a 4 horse race, Cisco (Stratacom), Lucent (Ascend, Cascade, Nexabit?), Alcatel (Newbridge), Nortel (Passport).

Marconi only ever had a strong market position with ATM in the SP with the ISP's (Who no longer use ATM in their cores) and with Sprint (Who have an interesting ION architecture which is not typical). Marconi have possibly built a great SP ATM switch at last, problem is they have no market share (3% ish), no account control in the data side of the SP (Transmission and Voice a bit) and their product will not integrate well with the existing deployed ATM kit and management systems (The other vendors will make sure of that).
Marconi are toast for sure. How about ATM the movie?

Of the other 4.... They each own customers and they each have a strategy for upgrading their customers networks. I think Lucent and Alcatel are in good shape, not sure about Cisco and Nortel. Equipe.... I'll just provision another lambda, do you want fries with that? yea right.

Now about this hybrid ATM/IP/MPLS switching market......

Well I think that the first point is that no SP that I am aware of actually builds real hybrid networks. They have entirely separate IP cores and Frame/ATM cores, or else some sort of overlay where one provides service to the other.

Personally I think they are bonkers, they are trying to mix together Oil and Water.

I think that most interesting point to make here is related to the customer service. Customers all want to use IP for data networking applications and Customers all want to get value for money in terms of bandwidth and service.

The problem for the operators is that IP networks are not really architected according to the old telco model of 90 cents a minute, they are more an all you can eat type buffet.

This is the real circle that operators need to square. How to deliver gobs of bandwidth for pennies and still afford a CEO earning millions!

I would suggest that they can and are addressing this in one of two ways.

Try to maintain as much control as possible over the access network infrastructure (This is why the SP's paid silly money for 3G licenses) to stop Customers from being able to go elsewhere.

If they can maintain control, then milk their cashcows for all they are worth (Frame) and in parallel introduce new higher value services that lock customers in to them (IP VPN 2547).

They will invest what they have to to keep milking their cashcows (Larger ATM core switches and higher speed links, plus new edge technologies) but no more than is necessary to keep the gravy train rolling.

In parallel they will invest in their IP service networks to add new high value and sticky services, like bundles of high speed Internet access and new VPN services (L3 and L2 VPN, IPSEC)

The communications services that I want to buy in the 21st century are Voice (Mobile and Fixed) and now IP/Internet access. I want wireless telephony and wireless data in my home and office and everywhere in between, oh and I don't want to pay more than $50 a month!

The SP's need to figure out how best to deliver these services and what technologies they need to do it in a cost effective way. This is a non-trivial task, but personally I think they they would be well advised to avoid trying to integrate ATM with IP, life is going to be quite complicated enough without trying to mix these two in the networks.

IMHO hybrid ATM/IP/MPLS control planes will result in great complexity and cost, and will not in fact deliver the benefit of convergence in terms of a single and therefore (Surely!) lower cost network infrastructure.

Perhaps this is not what operators want to hear or believe, and it may not be what the marketing folks at the ATM vendors want to hear either.

flanker 12/4/2012 | 7:24:34 PM
re: Switch Vendors to Tackle Core Routing Well I think that the first point is that no SP that I am aware of actually builds real hybrid networks. They have entirely separate IP cores and Frame/ATM cores, or else some sort of overlay where one provides service to the other.

I have heard this repeated before by many posters, but the reality is that most IXCs run IP over layer 2 virtual circuits. UUNET and Cable and Wireless, arguably the two largest IP carriers, both run IP over ATM and are migrating to MPLS (Dont know if that's IP Over ATM + MPLS or IP + MPLS). Level3 is using MPLS, even though they touted the network as an IP backbone originally.

By hybrid network perhaps you mean ATM and IP either-or provisionable between nodes, but I am not sure that this is the driver for these multi service platforms (then again, for a carrier using up valuable rack space for routers and ATM nodes, maybe it is). They are migration platforms. I think within 5 years ATMs functionality wil have been absorbed to into one of the emerging protocols.

Mr Moochin 12/4/2012 | 7:24:33 PM
re: Switch Vendors to Tackle Core Routing Although this has been touched on isn't one of the main reasons to purchase these big ATM/IP boxes to enable existing telco's/ISP's using ATM to upgrade their existing core to STM-16/64 from the traditional STM-1/4? Thereby saving the massive expense of installing a large IP network ala Juniper M160/Cisco OSR?

Especially in the existing financial climate this must prove a massively desirable option?
lr_monger 12/4/2012 | 7:24:33 PM
re: Switch Vendors to Tackle Core Routing Equipe has a great play as a point product and they will win because of it and the other factors about them (they have been there and done it).

The other ATM switch vendors claiming to have added IP routing to their products are merely putting lipstick on their pigs.

lr monger

elbowjames 12/4/2012 | 7:24:32 PM
re: Switch Vendors to Tackle Core Routing Lipstick on Pigs. very ggod. I like it a will use it. No copyright I assume?
Packet Man 12/4/2012 | 7:24:29 PM
re: Switch Vendors to Tackle Core Routing The Marconi (FORE) ASX4000 used to be a pure ATM only switch. I kept hearing from the sales guys and account managers that ATM is the only way to go, "why buy routers"? Well hmmmm looks like you can buy IP router blades for this thing.

I have no idea about the BXR-48000 but it looks like a "anything over anything" box.
lu-alum 12/4/2012 | 7:24:28 PM
re: Switch Vendors to Tackle Core Routing Hmmm...wonder where RouteThis works
<<   <   Page 2 / 6   >   >>
Sign In