Intelligent Bandwidth Management for the Intelligent InfrastructureTM

The ATM switch network modules provide the physical interfaces for the ForeRunner® ASX-200BX, ForeRunner ASX-1000, TNXTM-1100 and TNX-210 ATM switches. The ATM network modules provide LAN and WAN connectivity to other ATM switches, ATM-capable desktop computers, servers, routers, LAN switches, ATM concentrators and wide area ATM services.

Additionally, the VoicePlusTM network module supports interworking between TDM T-1/E-1 and ATM environments, and the FramePlusTM network module offers standards-based internetworking between ATM and Frame Relay or ATM frame-based UNI (FUNI) services.

The network modules feature industry-standard interfaces ranging from T-1/E-1 to OC-12c/STM-4c supporting fiber, coax and copper media. This variety of interfaces gives the ASX and TNX switch families the flexibility to support a wide range of enterprise and service provider LAN and WAN applications.

Sophisticated per-VC bandwidth and congestion management techniques ensure that contracted service levels are satisfied for each connection. Per-VC buffer management effectively isolates users from one another protecting each user's service level from being impacted by other connections.

The network modules adhere to the latest standards for traffic management, physical connectivity, facility maintenance, diagnostic loopbacks, etc., ensuring interoperability with standards-based ATM services and equipment.

The family of ATM switch network modules are interchangeable across the TNX and ForeRunner ASX product lines, maximizing investment protection as network modules are moved from one class of switch to another to accommodate network expansion and new service level requirements.

System capacity is easily grown by simply adding a network module to one of the modular ATM switches. The modules are hot swappable, supporting in-service expansion and replacement, reducing maintenance downtime.

Smart Buffers and Dynamic Buffer Management

Smart buffer management techniques maximize the effectiveness of the modules' large output buffers (up to 65,536 cells) by dynamically allocating bandwidth between connections. When compared to statically-managed buffers, smart buffers achieve a 200 to 300% increase in effective buffer capacity, as shown below. Smart buffer management provides the flexibility to assign buffers to connections on an as-needed basis, maintaining the Quality-of-Service (QoS) of CBR and VBR connections, while accommodating traffic bursts from UBR and ABR connections.


Dynamic vs. Static Buffers

Figure 1: Comparison of Dynamic vs. Static Buffers:

This diagram represents a two-port Network Module with ports labeled as 'A' and 'B.'
(The figure on the right shows statically-allocated buffers.)
Static buffers do not have the flexibility to reassign buffer space on an as-needed basis. Therefore, they drop several cells from port 'A' even though buffer space is unused and rigidly assigned to port 'B.'
(The figure on the left shows dynamically-allocated buffers.)
The dynamically-allocated buffers can support both 'A' and 'B' traffic because they have the flexibility to assign buffers to each port on an "as-needed" basis.


Per-VC Shaping and Scheduling

The Series D line of network modules features per-VC shaping and scheduling in which outgoing cell streams are paced by embedded hardware and software logic. This ensures that each VC and VP connection conforms to its specified service contract. This capability is particularly important when connecting to public ATM services where policing can result in lost traffic because of non-conformance to established service contracts.

In particular, unshaped connections are susceptible to this policing action in which traffic behavior such as cell clumping (e.g., which can result from the statistical multiplexing of many connections) can result in a violation of certain QoS parameters.


Per VC Shaping and Scheduling

Figure 2: Per-VC Shaping and Scheduling Illustration


Frame Discard

Frame Discard improves data throughput by intelligently discarding all the cells from selected AAL-5 packets instead of dropping random cells from multiple packets. This smart discarding dramatically reduces the number of packets that must be retransmitted and increases data goodput (i.e., the number of successfully transmitted packets).

Two integrated Frame Discard techniques, Partial Packet Discard (PPD) and Early Packet Discard (EPD, work together to maximize goodput by localizing the impact of port congestion to as few packets as possible.


Frame Discard Comparison.

Figure 3. Frame Discard Comparison:

This illustration is based on actual testing that shows how these techniques improve goodput on congested ports by as much as 75%.


Explict Rate ABR

With ForeThought Release 5.2, Series D network modules support the Explicit Rate ABR service category specified in the ATM Forum Traffic Management Specification 4.0.

Explicit Rate ABR goes beyond earlier ABR flow control mechanisms by not only indicating that there is congestion, but also specifying the rates at which the traffic source must operate to avoid network congestion. This allows ATM traffic sources to generate the optimal amount of traffic, virtually eliminating congestion on the most heavily utilized ATM links, such as inter-switch links and connections to network servers.

World-Class Performance

These advanced traffic management features set a new standard for ATM network performance, enabling network operators to deploy scalable, high capacity ATM networks providing efficient network utilization while supporting fair, deterministic service to hundreds of thousands of end users and applications.