x
Routing

Foundry Claims Core Crown

Now that Foundry Networks Inc. (Nasdaq: FDRY) is back to selling carrier routers, it's getting bold. The company says its newest edition, being released today, can outdo Cisco Systems Inc. (Nasdaq: CSCO) and Juniper Networks Inc. (NYSE: JNPR) in the trickiest of markets -- namely, the network core.

Yes, that means Foundry says it's beaten out Cisco's CRS-1 and Juniper's T640 -- at least when it comes to what's available in a single box. Today, Foundry is unveiling the XMR 32000, a core router taking up half of one 7-foot telecom rack, claiming it's got something as powerful as other core routers but more compact.

The 32-slot XMR 32000 can sport 128 ports of 10-Gbit/s Ethernet, Foundry claims. The half-rack version of Cisco's CRS-1, by contrast, could do only 64 ports of 10-Gbit/s Ethernet in theory (that's eight slots each filled with an 8-port 10-Gbit/s Ethernet card). The T640 can only carry 32 ports of 10-Gbit/s Ethernet, according to Juniper's Website.

That kind of compactness matters to carriers, Foundry claims. "The industry is very concerned about power consumption and very concerned about cooling issues," says Ken Cheng, vice president of what Foundry calls the "high-value business unit." (We're assured that the other business units aren't considered "low-value.")

Foundry so far lacks the multichassis capability of the CRS-1 and T640, and it also lacks support for Sonet OC192 or OC768, or the SDH equivalents. So, it's not as if Foundry has made other core routers obsolete.

But the XMR 32000 shows how a company can benefit just by having newer chip technology. The T640 is relatively old, having been announced in 2002, and Cisco famously spent $500 million and several years cooking up the CRS-1. (See Juniper Goes Terabit With the T640 and Cisco Unveils the HFR.) By starting later, Foundry took advantage of more advanced semiconductor processes to produce what it says is a comparable box that's cheaper -- $90,000 for a base XMR 32000, versus the $450,000 price tag initially placed on the CRS-1 -- and capable of some nifty future stuff, such as 100-Gbit/s Ethernet support.

"Companies like Foundry create products like this on a much faster development cycle than Cisco and Juniper," writes Mark Seery, an analyst with Ovum RHK Inc. , in an email to Light Reading.

Cisco declined comment, citing a policy to avoid talking about competitors. Juniper wouldn't comment on the XMR 32000, but a spokeswoman notes it's not easy for a newcomer to win big core-router contracts. "Every time you see a customer like MSN or China Telecom Corp. Ltd. (NYSE: CHA) make a T640 or a TX decision, rest assured they are endorsing the product strategy for many years to come," she writes via email.

Foundry found this out in 2001, when the company launched a core router and got pummeled. (See Judgment Day for Foundry Core Router and Foundry Retreats from the Core.) Foundry backed off from the service provider market as a result of the downturn, moving instead to the healthier enterprise market.

Foundry has been paying attention to service providers again, though, and early in 2005, the company re-engaged the carrier-network core, introducing the IMR router line. While IMRs are still available, they've been usurped by the XMR family, which Foundry introduced at last year's Supercomm show. Separately, the company has extended the XMR line into the metro with the NetIron MLX variant. (See Foundry Strikes at the Core, Foundry Boosts Metro MPLS, and Foundry Reinforces Metro.)

But Foundry has to face the same black cloud it did the first time: "How many successful startups have we had in the core router space outside of Cisco and Juniper?" asks Sam Wilson, an analyst with JMP Securities .

Well, that would be darn near zero -- how near depends on whether one counts Avici Systems Inc. (Nasdaq: AVCI; Frankfurt: BVC7) as a success. (See Avici Downsizes to Survive and Avici Soars on Q1 Numbers.) The problem is that a core router represents a change of philosophy for a carrier; core upgrades don't come around often, and when they do, they tend to go to trusted vendors.

Plenty of core-router firms have failed, slipped into obscurity, or gotten gobbled up by Cisco. (See No Tomorrow for Chiaro and Cisco to Pay $89M for Procket Assets.)

Even Foundry's bonus points for compactness aren't swaying some analysts, including Wilson -- who points out that boatloads of hot technology couldn't save Procket Networks. He lauds Foundry's push into service provider networks, given the money to be found there; it's just that core routers are "a tough, tough market," he says.

"I certainly wouldn't expect [Foundry's XMR] to emerge as an immediate contributor in terms of going into a really large, live production network," says John Mark Duncan, an analyst with Pacific Growth Equities Inc.

Naturally, Foundry disputes that. "I've seem tremendous interest in the XMR from Tier 1 service providers, the people most interested in core routing," Cheng says. "They are up for another round of investment."

Still, the more likely targets for Foundry might include "wholesale IP transport providers, Internet exchange carriers, data centers, and high-performance computing environments," Seery suggests. "If Foundry can gain traction with smaller players and mature the software, they might be worth watching when the large carriers move to all-Ethernet cores."

Seery admits that move might be delayed, especially if the Institute of Electrical and Electronics Engineers Inc. (IEEE) decides against doing a 40-Gbit/s Ethernet standard. Still, should carriers choose the Ethernet path, creating a simpler network core, "then Cisco and Juniper will need to keep a close eye on products like the XMR 32000," he says.

— Craig Matsumoto, Senior Editor, Light Reading

Page 1 / 5   >   >>
optodoofus 12/5/2012 | 3:55:32 AM
re: Foundry Claims Core Crown "The industry is very concerned about power consumption and very concerned about cooling issues," says Ken Cheng.

Indeed they are. But it seems to me that cramming twice as much stuff in a half rack is not a formula for reducing heat dissipation issues. In fact, the reverse is quite likely to be true.

optodoofus
Pete Baldwin 12/5/2012 | 3:55:32 AM
re: Foundry Claims Core Crown But it seems to me that cramming twice as much stuff in a half rack is not a formula for reducing heat dissipation issues. In fact, the reverse is quite likely to be true.

A lot of the secret sauce to Foundry's box seems to be in plain old chip shrinks. In which case, it's believable to me that they're running cooler.

As always, we'd welcome arguments to the contrary...
Pete Baldwin 12/5/2012 | 3:55:32 AM
re: Foundry Claims Core Crown OK, I admit -- i picked that subject line just to get attention.

But I'd invite people to have some fun predicting Foundry's future in the core. Maybe Cisco will pre-emptively acquire the whole company (KIDDING.)

I can believe that Foundry's getting some advantage from having newer chips. But obviously, the others can catch up on that front. Mark Seery says Juniper is likely to double the T640's density this year, for example -- seems like an upgrade that's well overdue.
willroute4food 12/5/2012 | 3:55:31 AM
re: Foundry Claims Core Crown Although their density and time to market are very impressive, I haven't heard much regarding improvements to their software - ISSU, resiliency, overall improvements in stabilty, etc... What about track record? How many Tier 1 backbones are built with Foundry gear today? I'll give you 3 guesses, but I think you'll only need one.

These are also very important considerations when evaluating a core box because of the sheer volume of traffic and degree of impact when a device carrying 128 - 10GE interfaces bounces. **There goes New England** I am sure many carriers are interested. They would be remiss for not looking at a box with the spec sheet this one has. But, density is only a single data point.

Don't get me wrong, I dont want to come across as bashing thier accomplishments. This box certainly provides them a good base to improve upon and the density is also impressive. It'll be interesting to see whether they can capitalize on it and gain momentum in the core or not. My guess is they stay in the datacenter.

Just one man's opinion,

WRFF
twill009 12/5/2012 | 3:55:30 AM
re: Foundry Claims Core Crown One might have thought we were past all the classic 'speeds and feeds' marketing hyperbole, but then again we are talking about Foundry. Juniper found success because it realized the OS was as important to customers as the hardware. It is not clear at this point whether Foundry is carrier-grade on this score.

If anything, this announcement shows that equipment companies smell an upturn in carrier spending and are willing to expend some effort to get back into the space. Let's hope it doesn't rest on false hopes.
mtb826 12/5/2012 | 3:55:30 AM
re: Foundry Claims Core Crown Core routing goes beyond the chipset, sw has to be feature rich and extremely stable. Just ask Cisco, it's the reason so many CRSs are still sitting in the lab. L3 routing has never been FoundryGÇÖs strength and their reputation in the Carrier space reflects that history. Overcoming their past will be one of many challenges they face to make the short list.

But does Foundry's push back into the IP Carrier space make them an attractive take over target for NT or Ericsson?
photon2 12/5/2012 | 3:55:29 AM
re: Foundry Claims Core Crown Good point Turin....there are two markets here: One at the Internet core of cores where in some cases L2 switching is used...and, Force10 has a lot of that market. Second market is Internet routing with BGP...HELLO, BGP...it takes years to get this right. If Foundry can't produce a OC-192 interface that Juniper and Cisco have had for years, at what level is their BGP development? Put it in the "Ethernet routers" category and watch Cisco and Juniper's market share remain unchanged. Too little, too late here is my take.
P2
turing 12/5/2012 | 3:55:29 AM
re: Foundry Claims Core Crown Yeah they're nuts.
If it's all 10Gig Ethernet and no oc-48/192c at least, then it's not a core router.
If it runs Foundry software, then no carrier would consider it a core router (or even a bgp router).
It's going to take a lot more than density and capex savings to make a carrier needing that type of capacity in the core change vendors.
I'm sure Ericsson and Lucatel know that too, if FDRY is thinking acquisition.

Why don't they go after the Force-10 market instead?
reoptic 12/5/2012 | 3:55:29 AM
re: Foundry Claims Core Crown After so many years of covering the core market would think LR would be even more circumspect. The core market first and foremost is not about speeds and feeds, it is about having stable and proven compatible software that runs the right feature set. Foundry is a long way from this. Also, a broader set of line interfaces needs to be available that what foundry is offering, capable of running huge numbers of routes, at line rate, with ACLs active. Even if you can achieve that you need long prove-in cycle from carriers willing to bet their network on you. Then you have to come tremendous economics favoring incumbents, not to mention aggressive pricing they offer to keep you out, with bundled tie-ins to other products they provide. Given all these factors predict Foundry will have minimal success other than some all-Ethernet oriented carriers such as Level 3.
Honestly 12/5/2012 | 3:55:28 AM
re: Foundry Claims Core Crown As for hardware, there are 10 gig ethernet solutions from Fujitsu that can hammer Foundry on a price per port/scale basis and CSCO (IOX) and Juniper (Junos) understand that it takes a carrier two years to test core software. Bobby, better play with the IXC's and fringe markets. As for HPC, you will get your lunch eaten, check out Myricom and other specialists in that space that scale to huge places with mature VLSI and SW. Stay with what you do well and pick off niche's. If you think you can be more in the core call Procket investors, that was the killer box and what happened.?
Page 1 / 5   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE