Shared Traffic Manager for Network Egress Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In network devices, the conventional approach of having separate traffic managers for each egress block leads to inefficient resource utilization, as resources are not fully utilized during peak traffic periods in one block while other blocks may have idle resources, resulting in data units being dropped due to resource limitations.
Innovation Solution
Implementing a shared traffic manager across multiple egress blocks, where resources are pooled together, allowing for increased capacity during peak traffic periods and enabling both dedicated and shared traffic management resources to coexist, optimizing buffer usage through read instruction queuing and caching mechanisms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If separate traffic managers are used for each egress block, then each block has dedicated resource allocation, but resource utilization efficiency decreases during peak traffic periods
Solution Approach 1:
Multiple traffic managers are merged into a single shared traffic manager that serves multiple egress blocks. The shared traffic manager pools buffer resources and read instruction queue resources across all egress blocks, allowing resources to be allocated dynamically based on actual demand rather than being siloed in dedicated per-block managers. This merging resolves the contradiction by enabling both reliable resource availability through pooling and improved utilization efficiency through shared access.
Solution Approach 2:
The shared traffic manager performs multiple functions that were previously distributed across separate traffic managers. It manages buffers and read instruction queues for multiple egress blocks simultaneously, making the system more universal. This multi-functionality allows a single traffic manager to handle peak traffic demands from any egress block while maintaining resource allocation flexibility, thereby improving both reliability and productivity.
2Ease of manufacture
If dedicated traffic manager resources are allocated to each egress block, then resource allocation is simple, but data units are dropped during peak traffic due to resource limitations
Solution Approach 1:
The patent merges multiple dedicated traffic manager resources into a single shared traffic manager with pooled buffers and shared read instruction queues. This consolidation maintains allocation simplicity through a unified resource pool while eliminating data unit drops by allowing resources to be dynamically allocated across egress blocks based on actual traffic demands, thereby improving reliability without sacrificing ease of resource management.
3Stability of the object's composition
If separate traffic managers are used for each egress block, then resource isolation is maintained, but overall system capacity is limited during peak periods
Solution Approach 1:
The shared traffic manager merges resources from multiple egress blocks into a unified pool while maintaining logical separation through read instruction queues associated with each egress block. This approach preserves the stability of resource isolation at the block level while achieving higher system capacity through inter-block resource sharing. The pooled buffers can be allocated to any egress block needing capacity during peak periods.
Solution Approach 2:
The patent introduces a new dimension of resource management by moving from per-block resource allocation to a system-wide pooled allocation model. Read instruction queues maintain egress block identification, allowing the system to track and manage resources at the block level while pooling physical buffer resources across all blocks. This dimensional shift enables both isolation stability and increased system capacity simultaneously.
4Reliability
If buffer resources are increased for each egress block, then data unit drops are reduced, but hardware complexity and cost increase
Solution Approach 1:
The shared traffic manager merges buffer resources across multiple egress blocks into a single pooled resource pool. This eliminates the need for each egress block to have its own dedicated buffer allocation, reducing overall hardware requirements. The pooled buffers are accessed through shared read instruction queues, maintaining data unit retention reliability while significantly reducing the total buffer capacity hardware must support compared to dedicated per-block buffers.
Data Source
AI summary
A traffic manager is shared amongst two or more egress blocks of a network device, thereby allowing traffic management resources to be shared between the egress blocks. Schedulers within a traffic manager may generate and queue read instructions for reading buffered portions of data units that are ready to be sent to the egress blocks. The traffic manager may be configured to select a read instruction for a given buffer bank from the read instruction queues based on a scoring mechanism or other selection logic. To avoid sending too much data to an egress block during a given time slot, once a data unit portion has been read from the buffer, it may be temporarily stored in a shallow read data cache. Alternatively, a single, non-bank specific controller may determine all of the read instructions and write operations that should be executed in a given time slot.