EZchip Semiconductor Ltd. (NASDAQ:EZCH) has unveiled the NPS – Network Processor for Smart networks – a new family of network processor units (NPUs). The company says the product is “a revolutionary NPU that will enable the next generation of smart high-performance carrier and data-center equipment.”
Until now, network equipment vendors essentially had a choice of two types of processors: NPUs that provide fast Layer 2-3 processing based on complex microcode programming, and CPUs (single or multi-core) that provide flexible C-programmable Layer 4-7 processing but at the expense of reduced performance and power efficiency. Through its new architecture, NPS breaks the barriers imposed by traditional NPUs and CPUs, the company claims. It is designed to enable extremely high-performance C-programmable Layer 2-7 processing with an architecture that scales as more advanced services at higher speeds are required over time.
“We are bringing to market a new breed of network processor that is architected to address the next generation of smart high-performance carrier and data-center networks,” said Eli Fruchter, president and CEO of EZchip Technologies. “We are preserving NPU performance advantages, opening the Layer 4-7 markets for us, and doubling EZchip’s addressable market above and beyond serving edge routers. As for our main market of edge routers, NPS extends our competitive advantage.”
For carrier equipment, the NPS makes it possible to move advanced Layer 4-7 features from specialized services cards to common line cards, and to add new baseline features to the line card such as application recognition and IPSec VPNs. For data-center equipment, the NPS enables performance to be scaled to meet the required loads of the evolving data center, enabling greater Layer 2-7 capabilities within constrained power and space. It also can help data-center appliances as well as carrier equipment support line-rate performance for applications such as load balancing, firewall, and network virtualization via OpenFlow/software-defined networks, the company says.
EZchip says the NPS product family is uniquely designed for data-plane processing where packet processing and wire-speed forwarding prevail, in contrast to CPUs that are designed for general processing where management functions and control-plane processing prevail. Central to the NPS are the C-programmable Task Optimized Processors (CTOPs). These processing engines are designed specifically for data-plane processing. The optimized design allows the integration of 256 such processors, each with 16 threads, for a total of 4K virtual processing engines.
“We applaud EZchip for its bold move in developing the NPS, which truly redefines the network processor by stripping away traditional limitations,” said Bob Wheeler, senior analyst at analyst firm The Linley Group. “Only the market leader could develop such a ground-breaking product as the NPS while preserving its customers’ software investments by also extending the NP line.”
Initially, EZchip will deliver the NPS-400 and NPS-200 products with 400-Gbps and 200-Gbps throughput, respectively. Samples are planned for Q4 2013 and other derivative products will follow. EZchip customers for the NP-2/3/4/5 products will have the option to maintain the benefits of code portability and scale up with a future NP-6, or C-program their applications and benefit from the new capabilities of the NPS.
For more information on network processing ICs and suppliers, visit the Lightwave Buyers Guide.