In-Network Caching vs. Redundancy Elimination

TL;DR

This paper compares in-network caching and redundancy elimination techniques, demonstrating that caching architectures generally outperform RE solutions across diverse network scenarios.

Contribution

It provides a comprehensive comparison between in-network caching and RE, highlighting the advantages of caching architectures based on real network topology evaluations.

Findings

In-network caching outperforms RE solutions in various traffic conditions.

Caching architectures offer better scalability and efficiency.

RE techniques are less effective in dynamic network environments.

Abstract

Network-level Redundancy Elimination (RE) techniques have been proposed to reduce the amount of traffic in the Internet. and the costs of the WAN access in the Internet. RE middleboxes are usually placed in the network access gateways and strip off the repeated data from the packets. More recently, generic network-level caching architectures have been proposed as alternative to reduce the redundant data traffic in the network, presenting benefits and drawbacks compared to RE. In this paper, we compare a generic in-network caching architecture against state-of-the-art redundancy elimination (RE) solutions on real network topologies, presenting the advantages of each technique. Our results show that in-network caching architectures outperform state-of-the-art RE solutions across a wide range of traffic characteristics and parameters.