Brocade Boasts Bigger 10GE

Brocade Communications Systems Inc. (Nasdaq: BRCD) is announcing a couple of switching advancements Wednesday and while neither is all that radical, both show the vendor upping its game in key competitive areas.

Both products are likely to be highlights of Brocade's analyst day, being held Wednesday at its San Jose, Calif. headquarters.

Bigger fabric
For data-center fabrics, Brocade is introducing the VDX 8770 Switch, a larger, modular version of the VDXs already shipping. (Brocade has been referring to the new box as Project Mercury.)

Each blade can support 48 ports of 10Gbit/s Ethernet, for a total of 384 in the eight-slot configuration. Brocade says the box has switch-fabric capacity for all that traffic; in fact, the box is built with 100Gbit/s ports in mind, but those interfaces are in Brocade's plans for late 2013, says Premal Savla, a Brocade senior product manager.

Brocade's VDX, which the company claims has more than 700 customers, is competing with other data-center fabric options, among them, FabricPath from Cisco Systems Inc. (Nasdaq: CSCO) and the QFabric from Juniper Networks Inc. (NYSE: JNPR). ("Fabric" in this case refers to the data center's arrangement of switches, not the chip inside the switches.)

Brocade calls its fabric Brocade One, and like FabricPath, it's based on the Trill protocol. Avaya Inc. pitches a fabric based on another protocol called Shortest Path Bridging. And while Juniper hasn't detailed the innards of QFabric, there's recent speculation that it runs on a variation of Multiprotocol Label Switching (MPLS) .

For more

Down in the core
On the carrier side, Brocade is also announcing a move toward the dense Layer 2 "supercore" that some service providers are talking about.

The idea there is to place a massive Layer 2 switch in the core, often using MPLS label-switched paths (LSPs) in place of core-router connections, primarily to save money. Juniper's PTX addresses this market and Cisco is offering MPLS blades to let its CRS-3 core router handle this function as well.

To that end, Brocade on Wednesday is announcing a 24-port 10Gbit/Ethernet blade for its MLXe routers -- meaning its approach to the supercore resembles Cisco's more than Juniper's. It's a density leap for Brocade, which has only offered up to eight 10Gbit/s Ethernet ports per card for the MLXe. (A two-port 100Gbit/s card is available, too).

Brocade says the new card enables a maximum 768 10Gbit/s Ethernet ports per MLXe chassis. Juniper puts up similar numbers, as it offers a 24x10Gbit/s Ethernet card for the PTX and can likewise support 768 10Gbit/s ports on the larger version of that product, the PTX9000.

Cisco's MPLS blade for the CRS-3 supports 140Gbit/s, according to the company's data sheets.

For more

— Craig Matsumoto, Managing Editor, Light Reading

Houman0 12/5/2012 | 5:21:17 PM
re: Brocade Boasts Bigger 10GE

A fair comparison of alternatives in the core has to include the 7950 XRS, not just the historical status quo... The *80* 100GE ports or 800 10GE ports of the 7950 XRS, with L3 IP routing *and* MPLS LSR switching capabilities on the same hardware is the new standard. Brocade's 32 100GE and 768 10GE ports is certainly notable in comparison to PTX and CRS-3, where one achieves density at the expense of dumping routing functionality and the other maintains broad functionality (albeit on an array of different specific linecards) but suffers badly on density & physical plant considerations (space, power, weight...).

As you said, the announcement is a density leap for Brocade. I don't believe it's a leap for the industry, though. The true leap was made by ALU earlier this year with the 7950 XRS: delivering leaps in density (5x today's 100G), power efficiency (1/3 today's power draw), and versatility (cost-effective for MPLS LSR only *and* IP (P-routing & VPN) implementation options on a single platform, using a single consistent set of cards).

Houman Modarres

Director Marketing - Alcatel Lucent IPD


Pete Baldwin 12/5/2012 | 5:21:15 PM
re: Brocade Boasts Bigger 10GE

Thanks Houman. Wouldn't want to forget the XRS.

Pete Baldwin 12/5/2012 | 5:21:14 PM
re: Brocade Boasts Bigger 10GE

One interesting thing Brocade brought up was that they want 20-port 100GE cards eventually. But the interfaces are too big to allow that.

I would imagine AlcaLu's XRS and other companies' most recent designs are in a similar position, engineered for terabits that they can't get the interfaces for.

I just thought Brocade had an interesting point there, especially with ECOC coming up. Obviously, the module makers are working on smaller 100G (like the eventual CFP4 on the client side). Price will be an issue too, of course, but it sounds like the form factor has a ready market waiting.

patentchoi 12/5/2012 | 5:21:11 PM
re: Brocade Boasts Bigger 10GE

FabricPath, Q-Fabric, Brocade One, Avaya VENA Fabric etc are proprietary interconnects that are "based on open" standards. This is a lockin nightmare for customers.

The ones I haven't heard harping on a "Fabric" are Arista and Alcatel-Lucent. Maybe they will weave a fabric story pretty soon.

Sign In