Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Learning-Based Caching Mechanism for Edge Content Delivery

Hoda Torabi, Hamzeh Khazaei, Marin Litoiu

TL;DR

This work addresses edge caching in resource-constrained environments by proposing HR-Cache, a learning-based eviction framework guided by Hazard Rate Ordering (HRO) upper bounds. It reconstructs the hazard-rate sequence from a window of past requests using a kernel hazard estimator and trains a gradient-boosted decision tree to predict the cache-friendliness of incoming requests, prioritizing cache-averse items for eviction. Empirical results on four real-world traces and one synthetic trace show that HR-Cache yields 2.2–14.6% greater WAN traffic savings than LRU (≈9.7% on average) and outperforms state-of-the-art learning-based policies, with significantly lower prediction overhead due to batch inference. The approach demonstrates the value of combining non-parametric hazard-rate estimation with learning to adapt cache decisions to diverse workloads, and it outlines future work for distributed and hierarchical caching in production environments.

Abstract

With the advent of 5G networks and the rise of the Internet of Things (IoT), Content Delivery Networks (CDNs) are increasingly extending into the network edge. This shift introduces unique challenges, particularly due to the limited cache storage and the diverse request patterns at the edge. These edge environments can host traffic classes characterized by varied object-size distributions and object-access patterns. Such complexity makes it difficult for traditional caching strategies, which often rely on metrics like request frequency or time intervals, to be effective. Despite these complexities, the optimization of edge caching is crucial. Improved byte hit rates at the edge not only alleviate the load on the network backbone but also minimize operational costs and expedite content delivery to end-users. In this paper, we introduce HR-Cache, a comprehensive learning-based caching framework grounded in the principles of Hazard Rate (HR) ordering, a rule originally formulated to compute an upper bound on cache performance. HR-Cache leverages this rule to guide future object eviction decisions. It employs a lightweight machine learning model to learn from caching decisions made based on HR ordering, subsequently predicting the "cache-friendliness" of incoming requests. Objects deemed "cache-averse" are placed into cache as priority candidates for eviction. Through extensive experimentation, we demonstrate that HR-Cache not only consistently enhances byte hit rates compared to existing state-of-the-art methods but also achieves this with minimal prediction overhead. Our experimental results, using three real-world traces and one synthetic trace, indicate that HR-Cache consistently achieves 2.2-14.6% greater WAN traffic savings than LRU. It outperforms not only heuristic caching strategies but also the state-of-the-art learning-based algorithm.

A Learning-Based Caching Mechanism for Edge Content Delivery

TL;DR

This work addresses edge caching in resource-constrained environments by proposing HR-Cache, a learning-based eviction framework guided by Hazard Rate Ordering (HRO) upper bounds. It reconstructs the hazard-rate sequence from a window of past requests using a kernel hazard estimator and trains a gradient-boosted decision tree to predict the cache-friendliness of incoming requests, prioritizing cache-averse items for eviction. Empirical results on four real-world traces and one synthetic trace show that HR-Cache yields 2.2–14.6% greater WAN traffic savings than LRU (≈9.7% on average) and outperforms state-of-the-art learning-based policies, with significantly lower prediction overhead due to batch inference. The approach demonstrates the value of combining non-parametric hazard-rate estimation with learning to adapt cache decisions to diverse workloads, and it outlines future work for distributed and hierarchical caching in production environments.

Abstract

With the advent of 5G networks and the rise of the Internet of Things (IoT), Content Delivery Networks (CDNs) are increasingly extending into the network edge. This shift introduces unique challenges, particularly due to the limited cache storage and the diverse request patterns at the edge. These edge environments can host traffic classes characterized by varied object-size distributions and object-access patterns. Such complexity makes it difficult for traditional caching strategies, which often rely on metrics like request frequency or time intervals, to be effective. Despite these complexities, the optimization of edge caching is crucial. Improved byte hit rates at the edge not only alleviate the load on the network backbone but also minimize operational costs and expedite content delivery to end-users. In this paper, we introduce HR-Cache, a comprehensive learning-based caching framework grounded in the principles of Hazard Rate (HR) ordering, a rule originally formulated to compute an upper bound on cache performance. HR-Cache leverages this rule to guide future object eviction decisions. It employs a lightweight machine learning model to learn from caching decisions made based on HR ordering, subsequently predicting the "cache-friendliness" of incoming requests. Objects deemed "cache-averse" are placed into cache as priority candidates for eviction. Through extensive experimentation, we demonstrate that HR-Cache not only consistently enhances byte hit rates compared to existing state-of-the-art methods but also achieves this with minimal prediction overhead. Our experimental results, using three real-world traces and one synthetic trace, indicate that HR-Cache consistently achieves 2.2-14.6% greater WAN traffic savings than LRU. It outperforms not only heuristic caching strategies but also the state-of-the-art learning-based algorithm.
Paper Structure (20 sections, 3 equations, 6 figures, 5 tables, 1 algorithm)

This paper contains 20 sections, 3 equations, 6 figures, 5 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Background & Related Work
  3. Learning-based Caching
  4. Use of Machine Learning to Improve Performance
  5. HR-Cache
  6. Hazard Rate-Based Upper Bound
  7. Hazard Rate Function
  8. Calculating Hazard Rate
  9. Learning From HRO
  10. Training Data
  11. Features
  12. Training HR-Cache
  13. The HR-Cache Policy
  14. Experimental Evaluation
  15. Batched Predictions
  16. ...and 5 more sections

Figures (6)

  • Figure 1: Architecture Overview of HR-Cache
  • Figure 2: HR-E Upper Bound Comparisons and Hit Probabilities for LRU and Belady's Algorithms Across Three Cache Sizes.
  • Figure 3: Comparison of HR-Cache to State-of-the-Art Heuristic Caching Systems for the IBM trace.
  • Figure 4: WAN traffic reduction compared to LRU across various cache sizes for HR-Cache and seven leading algorithms. HR-Cache consistently achieves 2.2–14.6% greater WAN traffic savings than LRU, outperforming the SOA alternatives.
  • Figure 5: Comparison of Byte Miss Ratios for HR-Cache, the Best Performing Policy, and LRU
  • ...and 1 more figures