(Page 0 of 1)

The service router challenge: increasing enterprise network speed

Multiservice enterprise networks are moving more diverse, higher bandwidth traffic than ever before, requiring new ways to increase network performance. Multiservice networks need to deliver higher speed for bandwidth-hungry multimedia services including voice and video while securing those communications for users connected through remote login or over a local Wi-Fi network. Business-critical data including VoIP calls and low-priority data such as routine backups have vastly different performance requirements but traverse the same network, requiring prioritization for speed.

Service routers provide an integrated solution for these multiservice enterprise networks – prioritizing traffic as best-effort or mission-critical to help optimize application performance; providing traffic management and other quality-of-service (QoS) mechanisms to ensure low latency, line-speed throughput for mission-critical applications; and integrating encryption/decryption and firewall capabilities for security. With system capacities ranging from less than 10Gb/s to several terabytes per second, service routers can be used for a wide range of network applications including multiservice business networking services, residential service delivery and backhaul aggregation and gateways for mobile networks.

Smaller service routers are implemented in a 1-2 rack-unit form factor and include a single x86 multicore processor. However, their throughput is no longer adequate for many enterprise networks, where every connection is at least 1Gb/s. Larger routers have multiple x86 multicore processors, but are significantly more expensive.

One way to accelerate service routers to meet future enterprise network requirements is by using multiple high-performance x86 processors. However, this approach can dramatically increase system cost and power consumption.

Multicore processors accelerate multiservice enterprise networks

Another approach – coupling dedicated multicore processors with the x86 host processor – can deliver the highest acceleration for service routers in multiservice enterprise networks. The solution significantly increases service router performance while slashing hardware costs and power consumption per Gb/s. Avago® Axxia® Enterprise Communication Processors integrate multiple high-performance processor cores, Ethernet switching and hardware acceleration optimized for security, traffic management and packet processing. Application development kits significantly reduce time to market for service routers and other networking applications.

Integrated multicore processors feature multiple high-performance ARM or PowerPC complexes with acceleration engines for specific networking functions. The Avago AXE4500 shown below features four ARM® cores with shared L2 and on-chip L3 caches. The coherent memory interconnect provides a low-latency, high-speed connection that maintains cache coherency among the ARM cores, L3 caches, external memory and acceleration engines.

The Axxia Communication Processor AXE4500 also features a deep packet engine to identify applications, a traffic manager to prioritize traffic, and a modular packet processor to support low-level network protocols including IP and legacy TDM. The Avago AXE4500 integrates 1/10GbE, Serial RapidIO and PCI Express® connectivity for direct connection to network ports and other processors.

Axxia Processor memory interconnect