Carrier Ethernet

"By 2011-2012, the majority of access and aggregation equipment being deployed by carriers around the world will be IP, Ethernet, and WDM, not SONET/SDH. […] Especially in 2008, we're seeing more carriers using Ethernet, and more carriers conducting interoperability tests of all sorts of Ethernet products for residential broadband, business connections, and mobile backhaul."

- Michael Howard, Principal Analyst and Co-Founder, Infonetics Research

The demand for increased bandwidth and the pressure to slash port costs in access, edge, and metro-area networks has triggered the proliferation of Carrier Ethernet infrastructure. In 2007, worldwide Carrier Ethernet equipment revenue increased 27% over the previous year, and is expected to continue its double-digit growth over the next 4 years [1]. While Carrier Ethernet Switch/Router (CESR) platforms try to integrate multiplicatively higher densities of Gigabit Ethernet and 10-Gigabit-Ethernet ports (see graph below), the challenge posed to the silicon at the heart of these platforms is to process the breadth of subscriber traffic and services at such elevated aggregate rates – and to deliver this unprecedented processing bandwidth with the fastest possible time-to-market.

The Configurable Switch Array (CSA) from Cswitch is uniquely architected to meet the bandwidth, configurability, time-to-market, and cost/integration requirements of current- and next-generation Carrier Ethernet platforms. The CSA can perform the full range of high-bandwidth Ethernet traffic processing functions commonly deployed in a stackable or chassis-based CESR platform: traffic aggregation and oversubscription, packet switching and classification, and traffic management. Moreover, the CSA enables Carrier Ethernet equipment vendors to differentiate their feature sets directly in hardware and absorb emerging standards with rapid design cycle times.

Port Aggregation and Oversubscription

• CSA Solution Capabilities
  • Up to 6 x 10GbE / 24 x GbE termination using CSA embedded MACs
  • IEEE 802.3ad Link aggregation
  • L2-L4 ingress pre-classification or traffic snooping using PacketParsers
  • Dynamically reconfigurable link oversubscription topologies
  • User-differentiated algorithms for packet discard / statistical multiplexing
  • Packet buffering using dedicated high-speed memory controllers
  • Full RMON, flow classification, and packet drop statistics
  • Packet interface using Interlaken, SPI-4.2, or XAUI-like proprietary switch i/f
  • Optional egress classification, traffic shaping & queuing

Packet Switching & Classification

• CSA Solution Capabilities
  • 100 – 120 Gbps Ethernet packet switching using high-bandwidth DCC crossbar fabric
  • MEF 10.1 compliance
  • Support for IP/MPLS, VPLS, Q-in-Q, MAC-in-MAC, PBB-TE, and OAM functionality
  • Multi-protocol encapsulation / de-encapsulation and tunneling support using PacketParsers and RAUs
  • CRC32 generation & checking in RAUs at up to 128 Gbps
  • Fully configurable ingress and egress classification sequence
    • User-programmed PacketParsers dictate full L2-L7 field extractions
    • High-speed lookups using RAUs (hashing) or internal BCAM/TCAM
    • Larger database lookups using external memory / TCAM control
    • Variable latency pipeline allowing conditional, recursive, or parallel lookups
  • Per-flow or per-port policing and queuing
  • Xoff and/or credit flow control
  • Packet interface using XAUI, Interlaken, SPI-4.2, or proprietary uplink interface

Traffic Management

• CSA Solution Capabilities
  • 10 to 100 Gbps packet buffer and hierarchical ingress/egress traffic manager
  • RLDRAM or DDR based packet buffering via high-speed memory controllers
    • Dynamic buffer memory allocation w/ bandwidth reservation
    • Efficient buffer management to minimize quantization & banking effects
  • Configurable flow-based scheduling/shaping hierarchy up to 5 levels
    • Up to 128K flows (using internal memory for queue state)
    • Up to 1M flows (using additional external memory for queue state)
    • DRR / WRR / WFQ arbitration with SP in parallel
    • Dual token bucket shapers w/ 64Kbps resolution
  • Congestion avoidance using 64 WRED profiles
  • Configurable # multicast groups for L2, L3, VLAN flooding and pseudo-wire
  • Echo kill support for real and virtual ports
  • Per flow statistics (packet and byte)
    • Enqueue / Dequeue / Discard
    • Packet interface MAC statistics
  • Packet interface using XAUI, Interlaken, or XAUI-like proprietary switch i/f

For additional information on CSA example benchmarks and reference designs for Carrier Ethernet applications, please contact Cswitch Sales.

[1] Infonetics Research: Metro Ethernet Equipment Worldwide Market Share and Forecasts, May 2008

BACK TO TOP