[Paper Review] Joint Cache Resource Allocation and Request Routing for In-network Caching Services

Weibo Chu, Mostafa Dehghan+3 more|arXiv (Cornell University)|Oct 31, 2017

Caching and Content Delivery24 references3 citations

TL;DR

This paper proposes a joint cache partitioning and content-oblivious request routing framework for multi-content provider (Multi-CP) multi-cache networks to maximize overall network utility. By leveraging biconvex optimization and decentralized online algorithms, it achieves high hit rates and efficient resource allocation while supporting fairness and adaptability to dynamic traffic.

ABSTRACT

In-network caching is recognized as an effective solution to offload content servers and the network. A cache service provider (SP) always has incentives to better utilize its cache resources by taking into account diverse roles that content providers (CPs) play, e.g., their business models, traffic characteristics, preferences. In this paper, we study the cache resource allocation problem in a Multi-Cache Multi-CP environment. We propose a cache partitioning approach, where each cache can be partitioned into slices with each slice dedicated to a content provider. We propose a content-oblivious request routing algorithm, to be used by individual caches, that optimizes the routing strategy for each CP. We associate with each content provider a utility that is a function of its content delivery performance, and formulate an optimization problem with the objective to maximize the sum of utilities over all content providers. We establish the biconvexity of the problem, and develop decentralized (online) algorithms based on convexity of the subproblem. The proposed model is further extended to bandwidth-constrained and minimum-delay scenarios, for which we prove fundamental properties, and develop efficient algorithms. Finally, we present numerical results to show the efficacy of our mechanism and the convergence of our algorithms.

Motivation & Objective

• Address the challenge of efficient cache resource allocation in multi-content provider (Multi-CP) multi-cache networks where providers have heterogeneous traffic, business models, and QoS requirements.
• Formulate a joint optimization problem that maximizes the weighted sum of utilities across all content providers, considering both cache partitioning and request routing decisions.
• Develop a scalable, decentralized, and adaptive online algorithm that dynamically adjusts cache allocation and routing in response to changing traffic patterns.
• Extend the framework to bandwidth-constrained and minimum-delay scenarios, demonstrating its generality and robustness under diverse network constraints.
• Introduce a unified utility-based framework that supports various fairness criteria (e.g., proportional, max-min fairness) through utility function selection.

Proposed method

• Propose a cache partitioning mechanism where each cache is divided into dedicated slices per content provider, reducing contention and enabling per-CP performance tuning.
• Design a content-oblivious request routing algorithm that routes requests to caches based on performance metrics, ensuring hit probability depends only on allocated cache size.
• Formulate the optimization problem as a biconvex program, enabling efficient solution via alternating convex optimization techniques.
• Develop decentralized (online) algorithms that update cache partitioning and routing based on real-time traffic statistics, with periodic reconfiguration triggered by significant changes.
• Introduce a two-layer adaptive algorithm: upper layer monitors connectivity and latency; lower layer tunes cache allocation per CP under fixed routing to adapt to traffic shifts.
• Utilize utility functions (e.g., β-fair, log) to embed fairness objectives into the optimization, enabling proportional or max-min fairness as special cases.

Experimental results

Research questions

• RQ1 How can cache resources be jointly allocated and routed across multiple caches and content providers to maximize overall network utility in a Multi-CP Multi-Cache environment?
• RQ2 What is the structural property of the joint cache allocation and routing problem that enables efficient optimization, and how can it be exploited algorithmically?
• RQ3 How can the proposed framework be extended to handle bandwidth constraints and minimum delay requirements in real-world deployments?
• RQ4 In what ways can fairness among content providers be systematically incorporated into the cache resource allocation process?
• RQ5 How can the system maintain high performance and convergence under dynamic, real-time traffic changes?

Key findings

• The joint cache allocation and request routing problem is biconvex, enabling efficient solution via alternating convex optimization with proven convergence.
• The content-oblivious routing scheme ensures that each content’s hit probability depends only on its allocated cache size, simplifying system design and analysis.
• The hit rate of each content provider is linearly proportional to the total traffic volume directed to its assigned caches, enabling predictable performance scaling.
• The proposed decentralized online algorithms converge effectively and adapt to changing network conditions, with performance validated through numerical results.
• The framework supports various fairness models (e.g., proportional, max-min) by selecting appropriate utility functions, demonstrating its generality and practical relevance.
• Extensions to bandwidth-constrained and delay-optimized scenarios preserve key structural properties and enable efficient algorithm design, confirming the model’s robustness and scalability.