 networkZONE Products for the week of October 1, 2007
Lightstorm Networks Says…
Purpose-Built Carrier Ethernet Silicon Becomes Reality
Industry veterans unveil the Brooklyn-10 at Carrier Ethernet World Congress in Geneva
Lightstorm Networks has announced the availability of its first product, the Brooklyn-10, a carrier-grade 20Gb Layer 2 switch with support for next generation carrier Ethernet services for PBB-TE (PBT), T-MPLS, and the latest OAM standards. It is the first Ethernet switch Application Specific Standard Product (ASSP) purpose-built for the carrier market and developed in partnership with leading carriers over several years.
The Brooklyn-10 offers the latest carrier Ethernet services at wire-rate speed without the requirement of complex NPU or FPGA programming -- an industry first. The time is right for purpose-built Ethernet silicon. Carrier-grade Ethernet is needed to leverage the cost and simplicity of Ethernet, but to be effectively utilized by carriers it must have the network scalability, QoS, monitoring and protection capabilities enabled by prior generations of technologies such as ATM and SONET.
Driven by emerging standards, carrier networks are now rapidly adopting a network model that utilizes new packet technologies -- such as Multi-protocol Label Switching (MPLS) and carrier Ethernet standards such as Provider Backbone Bridging (PBB) and PBB-TE (PBT) -- at the metro edge and thru the optical core. Combined with a robust silicon ecosystem, Lightstorm is designed to take Ethernet and MPLS further into the most demanding networks that require guaranteed service delivery, unprecedented scalability and high reliability.
“The Carrier Ethernet market requires significantly different features than the Enterprise market. The business models between the two segments are very different where the carriers look to use their networks to generate monthly recurring revenue and enterprise users generally view their network as a one-time purchase fixed asset. Lightstorm Networks has recognized this distinct difference and with the Brooklyn-10 chip has introduced an Ethernet switch in silicon which services the needs of the carriers for network scalability, OAM, QoS, and protection,” said Eve Griliches, Program Manager at IDC. “Lightstorm Networks introduction of the Brooklyn-10 is well timed to correspond to the new wave of designs that OEMs are undertaking to address the carriers need for Ethernet based services.”
To-date, industry solutions have forced the utilization of costly FPGA’s or NPU’s, or have relied on OEM developed ASIC’s. Another approach has been to utilize enterprise Ethernet silicon and leverage control plane processors or FPGA’s to implement basic carrier services. While this approach was the only low cost option for early adopters, it lacked scalability and robust functionality. With Lightstorm’s exclusive focus on the carrier market, its suite of silicon, software and hardware products enable the next wave of Ethernet services, providing OEMs and TEMs with the best building blocks for rapid market introduction of new products in the carrier Ethernet market.
“As the demand for carrier-grade Ethernet adoption continues to steadily gain traction -- Insight Research projects total US public Ethernet revenues will grow to over $5 billion by 2012 - carriers need a solution scalable and powerful enough to meet their needs,” said Bryan Campbell, CEO and Founder of Lightstorm Networks. “We take the ease-of-use approach from the hardware-driven enterprise Ethernet silicon market and apply it to a more service-driven model for the carrier market, ensuring significant cost, scalability and reuse advantages for our customers.”
Brooklyn-10
The Brooklyn-10 Carrier Ethernet Switch is a carrier-grade Layer 2 switch with an integrated 20G traffic manager. It delivers efficient and intelligent performance without the need for data path programming required by today’s complex NPUs. The Brooklyn 10 offers significant network scalability, the latest Metro Ethernet Forum defined QoS capabilities, OAM support and protection. It is a complete PBB, PBB TE, VPLS and VPWS solution for the carrier Ethernet market. Driving “Ethernomics” to the Carrier Market with its innovative architecture, Lightstorm’s Brooklyn-10 key features include:
- Complete 20Gb L2 Switch with MEF compliant Traffic Management • 192MB of external buffer memory • Support for latest carrier standards
- PBB, PBB-TE (PBT) • VPLS, VPWS, and H-VPLS • CFM (802.1ag & 802.3ah) for 3.3ms OAM & EFM services • Massive Network Scalability
- 80K VLANs • 16K Pseudowires
- 128K MAC Addresses • MEF 9 &14 compliant • Flexible port configurations with integrated MAC’s & SerDes
- 10 – 1GbE Client Ports
- 1 – SPI4.2 Network Port for connectivity to 10G MACs, Framers, or Fabrics
- Scalable topology support to connect multiple Brooklyyn-10’s in a single system
The Brooklyn-10 is complimented with a robust carrier Ethernet software suite which leverages a well defined carrier Ethernet API and includes development tools for rapid device configuration and register viewing. To aid in OEM time to market a complete hardware development environment is also available.
EN-Genius Says…
Whether or not it’s technically correct, the idea of building Ethernet switch silicon that caters to the specific needs of metro edge networks or connection to EoS and WDM transport backbones has a certain truthiness to it (see Paul’s recent editorial in acquisitionZONE for a nice definition of this new term). That’s why I was not surprised to see Lightstorm’s Brooklyn-10 carrier-class L-2 Ethernet switch does so, as well as several other chip-level products uncloaking this week at the Carrier Ethernet World Congress in Geneva, Switzerland. It gave me a real sense of déjà vu since this same concept was responsible for the rapid birth and death of a fair number of silicon startups during the late 1990s telecom boom. Given the industry’s past history, it would be easy to dismiss this switch as another product chasing an illusory market but between the fact that Metro Ethernet is well on its way to becoming a mainstream technology and how much networking silicon has matured in the past five years I think Lightstorm may have a chance.
 The concepts the chip is based on are sound. The Brooklyn-10 huge addressing capability (80 k VLAN addresses, 128 k MAC addresses) and large buffers (up to 192 M of external DDR) make lots of sense for a metro environment. Likewise, its Ethernet-based OA&M support which accommodates the emerging IEEE 802.1ag connectivity fault management (CFM) and 802.1ah first mile messaging services handle these necessary functions just above the MAC level, without having to trouble the host system or resorting to fussy, expensive ASICs. The switch compliance provider backbone bridging extensions (MAC-in-MAC) and pre-standards 8021.1qay that eliminates the traditional spanning tree also seems to make Lightstorm well-positioned to grab a nice chunk of the Carrier Ethernet equipment market.
When asked whether their the chip could accommodate any changes in the so-called pre-standards that might arise before they were finally approved, Lightstorm seemed confident that it had enough programmability to handle what they anticipated. They also said that if the chip cannot accommodate a particular change there a way to perform the desired functions by can copying or diverting packets to the control processor, based on either register settings or MAC tables. This same feature should help designers add custom functionality to help differentiate their products without having to resort to spinning an ASIC or FPGA. I also wonder a bit if this will help the Brooklyn-10 deal with the tendency of several large network equipment manufacturers (for example Brand C and Brand F) to fiddle with standards like these to add proprietary functionality (and lock out competitors) but that’s only speculation.
Lightstorm gets big points in my book for waiting until they had operational devices before they announced their product. They report that the alpha version of Brooklyn-10 is out of the fab and in the lab with no functional errata and that it’s passed several critical functionality tests (E-Line performance and multicast stress tests) with flying colors and that functional testing for PBB, PBB-TE, and VPLS is underway. The same tests indicate a worst-case power dissipation of around 12 W.
Despite the encouraging indications, I still had a few real concerns about whether Brooklyn-10 is a viable product. My main worry is the same one I have for most startups who roll out a single chip in a market that is increasingly-dependent on end-to-end solutions, and large ecosystems of support and design tools to ease the complexity of the design effort. It was good to hear that they’ve arranged with Cortina Systems to come up with reference designs that hitch their mappers, framers and 10G MACs to the Brooklyn-10 network-side interface. They’ve also worked with IP Infusion to supply a carrier-class Ethernet protocol stack that runs on any standard PC processor.
Teaming with relatively well-established networking players like Cortina may give them the cachet they need to gain credibility with risk-averse networking equipment makers and the "design bulk" they need to attract ODMs with lightly-staffed engineering departments. I do worry however, that, despite the fact that they are using dated '90's era architectures in their Metro Ethernet silicon, Broadcom and Marvell's marketing clout, large array of reference designs, and the market inertia that goes with being in so many existing designs will pose a steep challenge for Lightstorm and other challengers to the established pecking order.
Lightstorm gets a half-saltshaker added to its Vapor Index Rating because they are a relative unknown and another half-saltshaker because my briefers were not able to supply me with some of the critical details of Brooklyn-10's architecture and operation in time for this review (mostly due to conflicting travel schedules and time zones). Despite the high sodium content, I think these chips are worth checking out and that Lightstorm is a company to watch. I'll look forward to reviewing one of their upcoming products to see if the good vibes I'm getting will translate to a better score next time.
Brooklyn-10 with a hardware evaluation platform is sampling. Volume pricing will be $250.00.
Data Sheet
|
