Powering AdvancedTCA Platforms with Network Flow Processors
By Nabil Damouny, Netronome

Accelerating Service Deployments Via a New Micro-Architecture

Service providers face many challenges today. On the one hand, their industry is very capital-intensive and requires huge investments. On the other hand, they must quickly offer and bundle new differentiated products and services to create additional revenue sources and to match competitors’ offerings. They face competition from cable and wireless providers as well as Internet-based voice services. In turn, they are requesting faster turnaround, greater flexibility, higher performance, and lower cost from their equipment providers.

Powering AdvancedTCA Platforms with a Field-Upgradeable Micro-Architecture

AdvancedTCA has proven to be an effective way to standardize hardware deployment, allowing service providers to build systems from a variety of best-of-breed solutions involving multiple vendors. When coupled with a new micro-architecture, this approach has many key advantages, including the ability to deploy field-configurable line cards or blades that can be programmed for specific functions, such as deep packet inspection or security processing.

Achieving flexibility and field upgradeability at speeds at or above 10Gbps requires a new micro-architecture embedded (for cost reasons) in a commercially available processor. This processor must be closely coupled with an industry-standard, general-purpose CPU in an AdvancedTCA-based system. To achieve the high performance requirements of carrier equipment in edge, metro, and core applications, it must deal with flows, not packets. Furthermore, it should take advantage of the now widely available multicore CPUs by directing specific flows to specific cores for highest system and power efficiency. This micro-architecture must support Layers 2 through 7 of the OSI model, as some applications require switching and routing (Layers 2 and 3), while others require content processing (voice, video, or data) and application processing (email or instant messaging) (Layers 4 through 7). One hardware design combined with service-specific software loads can do the job. The result is products that can both be deployed quickly and also offer longevity in the market. Such products can be field-upgradeable to accommodate new or finalized standards and emerging services.

An example device meeting these requirements is Netronome Systems’ Network Flow Processor (NFP). With the NFP, equipment providers can quickly build products that help a service provider rapidly deploy new services. The NFP can be used in-line with incoming or outgoing traffic, or as an accelerator in a look-aside configuration. It supports full Layers 2 through 7 processing and is field-programmable. It classifies incoming packets to specific flows, based on packet header or payload, and then forwards them based on pre-specified filtering rules. Some flows must be serviced by the host CPU and, hence, are forwarded to a specific core for optimum performance and power efficiency.

To quickly meet market demands and garner revenue from new services, service providers should look for new micro-architectures that can accelerate deployments. These micro-architectures (available soon from select equipment providers) will include AdvancedTCA platforms that are high-performance, programmable, and field-upgradeable.

Nabil Damouny is Director of Strategic Marketing at Netronome. You can reach him at nabil.damouny@netronome.com.