Cache-Aware Service Contact Selection for CATS Traffic Steering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The IETF CATS framework lacks a cache element in its functional architecture, which may result in selecting a service contact instance that is not optimal due to the absence of considering cached content availability, leading to redundant computations.
Innovation Solution
Incorporating a cache element interface in the CATS framework by using cache metrics alongside network and compute metrics to prioritize cached content availability, thereby optimizing traffic steering and reducing redundant computations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If the C-PS selects service contact instances based only on network and compute metrics, then the selection process is simple and fast, but the response time increases and redundant computations occur due to ignoring cached content availability
Solution Approach 1:
The patent segments the service contact instance selection process into multiple evaluation dimensions: network metrics (bandwidth, latency), compute metrics (processing capability, storage capability), and cache metrics (cached content availability). This segmentation allows the system to independently evaluate each dimension and combine them for comprehensive optimization, resolving the contradiction by adding cache evaluation without completely redesigning the selection architecture
Solution Approach 2:
The patent implements preliminary caching of content at service contact instances before actual service requests arrive. By pre-caching content and evaluating cache availability as part of the selection criteria, the system prepares optimal service instances in advance, reducing response time when requests are actually made without adding complex real-time computation
2Productivity
If the C-PS evaluates only network and compute metrics, then the evaluation process is computationally efficient, but redundant computations occur at service contact instances due to ignoring cached content
Solution Approach 1:
The patent implements a feedback mechanism where service contact instances report their cache status, content availability, and performance metrics back to the C-PS. The C-PS uses this feedback information to make informed selection decisions, prioritizing instances with available cached content. This feedback loop enables the system to avoid redundant computations by selecting instances that already have the required content cached, thereby reducing energy waste while maintaining evaluation efficiency
Solution Approach 2:
The system performs preliminary evaluation of cache metrics and content availability before routing service requests. By assessing whether cached content exists at potential service instances in advance, the system可以避免 unnecessary computation and data retrieval, thus reducing energy consumption related to redundant processing while keeping the evaluation process efficient
3Productivity
If the CATS framework lacks a cache element, then the framework structure is simple, but traffic steering optimization is insufficient due to inability to prioritize cached content
Solution Approach 1:
The patent integrates cache evaluation functionality into the existing C-PS and service contact instance architecture, making these components multi-functional. The C-PS now simultaneously evaluates network metrics, compute metrics, and cache metrics, while service contact instances serve both as compute resources and as cache storage. This universal approach enhances traffic steering efficiency without requiring entirely separate cache management infrastructure
Solution Approach 2:
The patent adds cache metrics evaluation as a distinct functional component within the CATS framework, segmenting the overall evaluation process into network evaluation, compute evaluation, and cache evaluation modules. This segmentation allows the framework to maintain its original simple structure while incrementally adding cache-aware capabilities, thus improving traffic steering efficiency without dramatically increasing framework complexity
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In a computing-aware traffic steering (CATS) network, such as those conforming to an IETF proposal, a service contact instance for a requested service is selected based on at least cache metrics associated with data cached at one or more egress nodes of the CATS network. In some embodiments, the selection is also based on compute and network metrics. In one implementation, the selection is based only on cache metrics unless that selection is too costly in terms of compute and/or network load. In that case, the selection is based on the compute and network metrics. In this way, cache metrics are prioritized over compute and network metrics as long as the cache-based selection is not too costly.