Bandwidth Efficient Cache Selection and Content Advertisement

TL;DR

The paper addresses bandwidth-efficient cache content advertisement and selection in multi-cache networks where indicators trade accuracy for overhead. It introduces SALSA2, a scalable adaptive and learning-based algorithm that jointly estimates mis-indication probabilities and accounts for inter-cache dependencies, while dynamically scaling indicator advertisements with two modes: full-indicator and delta updates. A formal probabilistic model uses positive/negative exclusion probabilities $\\pi_j$ and $\\nu_j$, and defines the per-query service cost $\\phi_x(D)$ that combines per-cache costs and the miss penalty: $\\phi_x(D) = \sum_{j \in D} c_j + M \prod_{\substack{j \in D\\tilde{I}_j(x) = 1}} \pi_j \cdot \prod_{\substack{j \in D\\tilde{I}_j(x) = 0}} \nu_j$. The contributions include a taxonomy of existing approaches, the SALSA2 Exclusion Probability Estimation framework, the cache selection and indicator advertisement algorithms, and extensive simulations on real traces showing up to 84% bandwidth savings and near-optimal service costs, demonstrating SALSA2’s effectiveness for modern dynamic caching. The results indicate SALSA2 can substantially reduce communication overhead without compromising data availability, making it a robust solution for scalable caching in heterogeneous networks.

Abstract

Caching is extensively used in various networking environments to optimize performance by reducing latency, bandwidth, and energy consumption. To optimize performance, caches often advertise their content using indicators, which are data structures that trade space efficiency for accuracy. However, this tradeoff introduces the risk of false indications. Existing solutions for cache content advertisement and cache selection often lead to inefficiencies, failing to adapt to dynamic network conditions. This paper introduces SALSA2, a Scalable Adaptive and Learning-based Selection and Advertisement Algorithm, which addresses these limitations through a dynamic and adaptive approach. SALSA2 accurately estimates mis-indication probabilities by considering inter-cache dependencies and dynamically adjusts the size and frequency of indicator advertisements to minimize transmission overhead while maintaining high accuracy. Our extensive simulation study, conducted using a variety of real-world cache traces, demonstrates that SALSA2 achieves up to 84\% bandwidth savings compared to the state-of-the-art solution and close-to-optimal service cost in most scenarios. These results highlight SALSA2's effectiveness in enhancing cache management, making it a robust and versatile solution for modern networking challenges.

Figures (3)

• Figure 1: The exact instantaneous exclusion probabilities (fullKnow) and the estimations produced by HeCS$_{\mathop{\mathrm{FNA}}\nolimits}$FN_aware_ToN and SALSA2.
• Figure 2: Service cost and bandwidth consumption of the heterogeneous 3-caches baseline scenario for varying traces and miss penalty values. The costs are normalized w.r.t. hypothetical perfect indicator.
• Figure 3: Service cost and bandwidth consumption when increasing the number of caches. The costs are normalized w.r.t. hypothetical perfect indicator (Opt).