SDN Switch Cache Replacement Index for Latency Sensitivity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In software-defined networking (SDN), the decoupling of control and data planes introduces potential packet delay due to cache misses, which is problematic for latency-sensitive applications like voice packets, as traditional cache replacement strategies do not adequately consider latency sensitivity, usage recency, and frequency.
Innovation Solution
A cache replacement strategy that computes a replacement index (RI) for flow entries based on latency sensitivity, usage recency, and usage frequency, prioritizing latency-sensitive applications and updating frequency weights dynamically to improve cache hit rates and minimize cache misses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If traditional cache replacement strategies are used in SDN switches, then device complexity is reduced and ease of operation is maintained, but packet latency increases for latency-sensitive applications due to inadequate consideration of latency sensitivity, usage recency, and frequency
Solution Approach 1:
The patent applies parameter changes by introducing a composite replacement index that incorporates multiple parameters (latency sensitivity, usage recency, and usage frequency) to evaluate and select cache entries for replacement. This multi-parameter approach optimizes cache replacement decisions to reduce packet latency for latency-sensitive applications while maintaining systematic control over the increased complexity through structured index computation.
2Adaptability or versatility
If a centralized SDN controller is used to manage flow tables, then network customization and optimization are improved, but packet delay increases when cache misses occur requiring communication between control and data planes
Solution Approach 1:
The patent applies preliminary action by proactively computing and maintaining a replacement index for cache entries based on latency sensitivity, usage recency, and frequency before cache misses occur. This pre-computation enables the switch to quickly identify and replace appropriate cache entries without requiring real-time communication with the centralized SDN controller, thereby reducing packet delay while preserving the benefits of centralized control for flow table management.
3Productivity
If cache size is increased to reduce cache misses, then cache hit rate improves, but device resources are consumed and may not be efficiently utilized
Solution Approach 1:
The patent applies parameter changes by dynamically evaluating cache entries using a composite index that considers latency sensitivity, usage recency, and frequency. This enables more efficient utilization of existing cache resources by prioritizing the retention of entries that are most critical for reducing packet latency, thereby achieving higher effective cache hit rates without necessarily increasing the physical cache size and consuming additional device resources.
Data Source
AI summary
In a network switch of a software-defined network (SDN) architecture, a method for improving cache replacement (CR) efficiency implemented therein, the method comprising maintaining a flow table comprising a plurality of flow entries, computing a replacement index (RI) value for each of the plurality of flow entries resulting in a plurality of RI values, wherein computing an RI value for a flow entry is at least partially based on latency sensitivity of a forwarding path specified by the flow entry, and generating, based on at least some of the computed RI values, an eviction list comprising a number of list entries, wherein each of the list entries points to one of at least a fraction of the flow entries.