O-SMCB Switch Shared Crosspoint Buffers Multicast Throughput
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current packet switches face challenges in efficiently handling multicast traffic due to high memory demands and cost, as they are primarily designed for unicast traffic, leading to potential buffer exhaustion and increased costs for memory implementation.
Innovation Solution
The implementation of a shared-memory crosspoint-buffered switch architecture (O-SMCB) where crosspoint buffers are shared among output ports, along with a scheduling mechanism and flow control to prevent buffer underflow and overflow, allowing for efficient replication and switching of multicast packets with reduced memory requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional CICB switches are used for multicast traffic, then high throughput and switching performance are achieved, but memory requirements become prohibitively large and costly
Solution Approach 1:
The patent merges the buffering functions of multiple output ports into shared crosspoint buffers. Instead of each output port having dedicated buffers, the invention combines buffer resources across multiple outputs, allowing efficient multicast packet replication while reducing total memory requirements from N² to approximately N×F buffers where F is the average fanout factor.
Solution Approach 2:
The shared crosspoint buffers serve multiple functions: they act as input buffers for unicast traffic, replication buffers for multicast traffic, and output queues for multiple output ports simultaneously. This multi-functionality eliminates the need for separate dedicated buffers for each function, significantly reducing overall memory requirements.
2Reliability
If dedicated crosspoint buffers are allocated for each output port, then buffer exhaustion is avoided, but implementation cost becomes prohibitive
Solution Approach 1:
The patent combines previously dedicated buffers into shared resources. By merging buffer allocations across multiple output ports, the system maintains adequate buffer availability for each output while reducing total buffer count and implementation cost from N² dedicated buffers to approximately N×F shared buffers.
Solution Approach 2:
The invention changes the buffer allocation parameter from dedicated (1:1 mapping between outputs and buffers) to shared (many-to-many mapping). This parameter change transforms the buffer architecture from N² dedicated buffers to approximately N×F shared buffers, reducing cost while maintaining reliability through shared resource access protocols.
3Quantity of substance
If input buffered switches are used, then memory requirements are reduced, but throughput and switching efficiency are limited
Solution Approach 1:
The patent merges the advantages of input buffering (low memory requirements) with output buffering (high throughput) by implementing shared crosspoint buffers that are accessible by multiple outputs. This combination achieves throughput comparable to output-buffered switches while requiring only approximately N×F buffers instead of N², matching the memory efficiency of input-buffered switches.
Data Source
AI summary
Multicast traffic is expected to increase in packet networks, and therefore in switches and routers, by including broadcast and multimedia-on-demand services. Combined input-crosspoint buffered (CICB) switches can provide high performance under uniform multicast traffic. However this is often at the expense of N2 crosspoint buffers. An output-based shared-memory crosspoint-buffered (O-SMCB) packet switch is used where the crosspoint buffers are shared by two outputs and use no speedup. An embodiment of the proposed switch provides high performance under admissible uniform and non-uniform multicast traffic models while using 50% of the memory used in CICB switches that has dedicated buffers. Furthermore, the O-SMCB switch provides higher throughput than an existing SMCB switch where the buffers are shared by inputs.