Foundry Intros Next-Gen 10-Gig Ethernet

The fight for 10-Gigabit Ethernet leadership is heating up.

Foundry Networks Inc. (Nasdaq: FDRY) today introduced its next generation of 10-Gigabit Ethernet line cards at the ComNet tradeshow in Washington, D.C. (see Foundry Touts 10-Gig Interfaces). According to the release, the company has improved its density of 10 Gbit/s by offering two ports on a single line card. The company also said it would slash its price on 10-Gbit/s Ethernet by 36 percent.

Foundry’s move to improve density and reduce its pricing follows on the heels of a similar announcement from startup Force10 Networks Inc. Last week, Force10 announced it was slashing prices on its dual 10-Gbit/s Ethernet port line card by 44 percent (see Force10 Slashes 10-GigE Pricing).

The startup, which designed its E1200 switch specifically for 10-Gbit/s Ethernet, has challenged Foundry and other established vendors, like Cisco Systems Inc. (Nasdaq: CSCO) and Extreme Networks Inc. (Nasdaq: EXTR), with its performance claims and aggressive pricing. The startup so far has been the only vendor to demonstrate full line-rate packet forwarding at 10 Gbit/s, and currently it claims to have some of the least expensive 10-Gbit/s Ethernet interfaces on the market.

“Improving the density and the price point of the technology is important for the industry,” says Michael Howard, principal analyst and founder of Infonetics Research Inc. “Force10 took the lead with reducing the price, and it’s working. The others have to follow with price reductions, because their sales people are getting hit with it when they visit customers.”

But the price cut still doesn't cover certain performance issues in Foundry's 10-Gbit/s systems. The BigIron, NetIron, and JetCore FastIron chassis, which all support 10-Gbit/s line cards, do not actually forward packets at 10 Gbit/s. Because the backplane is limited in capacity, it only supports about 8 Gbit/s of throughput per line card. This means that a dual-port 10-Gbit/s Ethernet line card only has enough capacity to support 4 Gbit/s of throughput per port.

Chandra Kopparapu, director of product marketing at Foundry, argues that the dual-port 10-Gbit/s Ethernet line card can support line rate throughput in certain configurations where the backplane of the switch is not fully utilized. For example, he says that in a ring topology, the interfaces forward at line rate. In this configuration, one port acts as the ingress, while the other acts as the egress onto the ring. Because each line card utilizes local switching that avoids the backplane, it can forward packets at rates of 10 Gbit/s.

This is a likely application for service providers, which often use fiber from existing Sonet rings to deploy Ethernet in their metropolitan area networks. But some critics argue that most carriers don't have enough traffic demand in their metro rings to need a 10-Gbit/s Ethernet solution.

“Today, about 80 to 90 percent of the demand for 10-Gbit/s Ethernet is in the enterprise,” says Infonetics’ Howard. “Universities and grid computing researchers are the ones most interested in the technology right now. And most of those networks aren’t using ring configurations.”

Foundry’s Kopparapu agrees that service providers, particularly in the U.S., are not using 10-Gbit/s Ethernet yet, but he says that plenty of enterprises use ring topologies to connect campuses together. Kopparapu admits that Foundry needs to develop a next-generation chassis, and he says the company is working on one. But he argues that all this hype over line-rate performance is overblown.

“Sure, we have a new system in the works,” he says. “But I disagree with the presumption that our current product doesn’t compete. It may only have a backplane capacity of 8 Gbit/s, but that doesn’t mean that it is 20 percent less effective, or that it isn’t usable. Looking at just one aspect of performance is a superficial analysis.”

So how does Foundry’s latest 10-Gbit/s Ethernet line card stack up in terms of price? On a per port basis, with the 36 percent reduction, Foundry’s list price is about $38,500 per port. This includes $61,000 for the base module with two ports and $8,000 per port for the optics.

Compare this to Force10’s announcement last week. The startup is listing 10-Gbit/s Ethernet ports with 1,310 nanometer optics included for $31,000 or a street price of about $17,000, says Steven Mullaney vice president of marketing for Force10.

But comparing port prices alone is often deceiving. Kopparapu suggests comparing the cost of a total system. Below is an itemized comparison between the 15-slot BigIron 15000 chassis from Foundry and a 16-slot E1200 chassis from Force10. Both chassis are fully loaded with 14 line cards, each with two ports of 10-Gbit/s Ethernet interfaces.

Table 1:
Foundry BigIron 15000 Solution U.S. List Quantity Total
15-slot chassis $11,995 1 $11,995.00
Redundant Power Supplies $2,495 2 $4,990.00
Management Module $11,995 1 $11,995.00
2-port 10GbE (New 10GbE Line Cards) $60,995.00 14 $853,930.00
LR XENPAK (Long-Range 1310nm Optics) $7,995.00 28 $223,860.00
Force10 E1200 Solution U.S. List Quantity Total
16-slot E1200 chassis, (Bundled solution) -- 6 fan trays, 2 DC Power Entry Modules, 1 Route Processor Module (ED), 9 Switch Fabric Modules $110,000.00 1 $110,000.00
2-port 10-Gigabit Ethernet line card with 1310nm serial 10Km optics $62,000.00 14 $868,000.00
Source: Foundry Networks and Force10 Networks

This shows how the products differ in price structure. Foundry’s chassis is cheaper, coming in at a cost of about $29,000. Force10’s loaded chassis lists for $110,000. But in a fully loaded configuration, both units would cost about $1 million, with the Foundry system costing about $128,770 more for a fully loaded configuration of 10-Gbit/s Ethernet.

While most customers today probably wouldn't populate an entire chassis with 10-Gbit/s Ethernet interfaces, this comparison sheds light on the impact of the per port costs on deploying the technology.

“Even though Foundry’s solution may still be a bit higher than the competition, their announcement is still a big move for the industry,” says Howard. “There are a lot of companies -- enterprise and service provider -- interested in 10-Gbit/s Ethernet. And every time the prices are cut, more will try it out.”

— Marguerite Reardon, Senior Editor, Light Reading
Page 1 / 5   >   >>
valleyguy 12/5/2012 | 12:46:23 AM
re: Foundry Intros Next-Gen 10-Gig Ethernet No IOS story so you really should not have to sacrifice here. Why not just buy Catalyst 6500s? This thing is not as dense, it's not scalable and Foundry is nowhere near delivery of the next generation box which I hear will be much larger, more expensive and less dense that what this little startup has right now.

Buying Foundry gear makes no sense. Just don't get anything you can build on here. It's higher price for a platform that is gasping for air and dying on the vine. And you have a glorified LAN switch deep in the heart of your network vs. a box built to run carrier networks that is fully redundant (including power supply). Packet over SONET? Fuggettaboutit.

Time to short Foundry stock. You can't buy Force 10 stock so maybe buy CSCO with your FDRY profits (if you got in at $4). This is going to stunt Foundry's growth.

Their sales force is panicking. Marshal Eisenberg left to join Force 10 (senior marketing Director). Party's over Bobby. You just can't build a franchise on a $20M development budget.
metroshark 12/5/2012 | 12:46:20 AM
re: Foundry Intros Next-Gen 10-Gig Ethernet Neither Foundry nor Force10 products are near the price point that may start driving demand for 10 Gigabit Ethernet. Until the 10Gig pricing gets below $10K, IT managers do not have much of an incentive to install 10 Gig ports in their networks.

Look at the per port pricing for fully populated configurations from Foundry and Force10:

Foundry has a very oversubscribed box with 28 10GE ports for about $1.1M. That is about $40K/port. And this is for less than 5Gb/s throughput per port. Assuming 40% discount off the list, the street price is probably going to be around $24K/port.

Force10's price for a 28-port 10GE configuration is $980K. This is about $35K/port. Again, assuming 50% discount off list, the street price is around $21K/port.

metroshark 12/5/2012 | 12:46:20 AM
re: Foundry Intros Next-Gen 10-Gig Ethernet I think there is a typo on 3rd paragraph from the end. Foundry pricing for the new 10GE linecards should be $77K for two ports, not $77K per port.
PacketGuy 12/5/2012 | 12:46:17 AM
re: Foundry Intros Next-Gen 10-Gig Ethernet Looks like valleyguy is like most Valley Girls I know - lots of talk but no brains!

Short FDRY? Did he miss their preannouncement on Jan 9th? http://biz.yahoo.com/rb/030109.... FDRY moved the tech. market because of their strong Q4 performance - it was all over CNBC. Valleyguy, please short FDRY with all that money that you were saving for a new trailer!

Obviously, you work for cisco. Buy a Catalyst 6500? Give me a break. Your comments about FDRY redundancy, price, packet-over-SONET, and carrier networks were ALL inaccurate. Does posting to LightReading from your cisco competitive marketing sheet makes you feel like a stud? You're probably young, but in time you'll learn not to drink all the cisco Kool Aid.
raypeso 12/5/2012 | 12:46:16 AM
re: Foundry Intros Next-Gen 10-Gig Ethernet I don't work for Cisco but my company is deploying the 6509's, at least a couple of them and I would just like to see a side by side comparison on the prices of the Foundry and Cisco boxes.
I think utilization might have been a big part of our decision making. I don't think we'd even come close to using 10G Eternet. I look at utilization for customers all the time and the ones we sell Gig E ports to are never even close.
Supposedly we got a great deal from Cisco, I think we've all heard that before.
Iipoed 12/5/2012 | 12:46:14 AM
re: Foundry Intros Next-Gen 10-Gig Ethernet Keep in mind the following is written by a reitred sales guy.

You are first of all comparing two different architectures. The 6509 is reworked old technology. Their real performance, their use of resources within their architecture, their forwarding of packets. Of course they do have probably the most stable software out there. To do a price comparison makes no sense. It comes down to what John Chambers and Bobby Johnson want to make margin wise and how important of a customer Cisco thinks you are. But you really cannot do and apples to apples cost analysis unless you take each box into the lab, test each blade, power supply, interface etc. and verify they are equal.

It really is the old addage "no one ever lost their job buying Cisco" of course they said that about IBM, ATT etc.

What future applications, what growth, what about not having to upgrade management modules when new blades become available. There are so many intangibles. Bottom line is Cisco wins majority of their business because of FUD. And I do own alot of CSCO and I hope they make it back to $50 a share. Of course I hope FDRY makes it back to $100 a share. IMHO

turing 12/5/2012 | 12:46:13 AM
re: Foundry Intros Next-Gen 10-Gig Ethernet The Foundry guy claims it's ok to not support 10Gig across the backplane because on a ring design traffic comes in one port to the other. That's a load of FUD. Is the card an RPR card? Are you really telling me a majority of the traffic is going THROUGH the two 10Gbps ports on the foundry and not to all the other ports? Then that means it doesn't need 10Gbps (even in bursts) Why bother with 10Gbps interfaces on it then? Just bundle some 1Gbps interfaces.

If you have a 2-port 10Gig Foundry card in the network it's because you either
1) really need 2 of 10Gbps ports
2) want redundant ports (ala link aggregation) while benefiting from their aggregate throughput when not in the failure mode

(1) cannot be supported by the Foundry. (2) can be supported in the failure mode (although even then they only do 8Gbps), but you don't get the aggregate 20Gbps benefit when it's not in failure mode! So for less money you could buy the Force10 box that does all of that! (of course Force10's problem is no one would risk buying from them these days - their only hope is to give away enough units to get market share to sell themselves to a bigger company)

This new Foundry card is a dud. They need a new backplane/fabric.
chitgo 12/5/2012 | 12:46:11 AM
re: Foundry Intros Next-Gen 10-Gig Ethernet It is clear all the switching vendors want to rope in as much revenues as possible for early 10GE deployment. Nobody has a switch today (i am leaving Force10 out of this!!) to support true 10GE speeds/QoS etc..I know for sure Extreme is working on releasing a new switch to support true multiple 10GE blades offering Wire Speed performance. Must be the same with Foundry i guess. With CSCO, it's anybody's guess!!!
BobbyMax 12/5/2012 | 12:46:09 AM
re: Foundry Intros Next-Gen 10-Gig Ethernet There are a lot of items are to be considered before a product selection is made. I would just list a few:

1. Cendor's reputation
2. Issue an RFP to the relevant vendors
3. Performance data at various loads
4. Reliability, MTBF, etc.
5. Network Management and Fault Reporting
6. Recovert tome after failure
7. Reliability of access and long haul interfaces
8. Transmission medium supported
9. Quality of documentation ease of installation
10. Quality of product support
11. Conditional buying
12. Request for SLA and the software they use to monitor
13. Software and hardware upgrades. What they support, hot swappable.
14, etc., etc.

tmc1 12/5/2012 | 12:46:07 AM
re: Foundry Intros Next-Gen 10-Gig Ethernet lipoed and the rest,

6500 is re-worked old technology (1999 6000 series) but so is netiron (1999 bigiron series). Both chassis are from same era and the 6500 has been more extensible with new WAN modules, processor based cards and software.

Gotta love Chandra saying not to just count performance. He is correct of course but when FDRY had the performance they were the ones using that to kick cisco's ass. I guess performance only matters when you are the one that has it!

FDRY vs. F10 is a no brainer. It will take years for F10 to catch up to the enterprise features that FDRY has in their software just as it took years for FDRY to catch up to cisco. The barrier to entry in the switch and router market is reaching a point where almost no startup can enter it anymore. I would say after this generation it will be almost impossible and the cost and risk will be HUGE!!! No VC will want to fund it unless their is a large paradigm shift.

As for the 10-gigE cards... The RPR/MRP is a cool application but no one is going to use it. These cards will sell like ice cream at the north pole. they need a new chassis over there in Alviso.
Page 1 / 5   >   >>
Sign In