Optimal Caching for Low Latency in Distributed Coded Storage Systems

TL;DR

This paper introduces caching strategies to reduce data access latency in geo-distributed erasure-coded storage systems, validated through experiments on cloud storage, balancing latency and computational efficiency.

Contribution

It proposes both offline and online caching schemes tailored for distributed coded storage, optimizing latency based on data popularity and network conditions.

Findings

Online caching approximates optimal solutions effectively.

Caching reduces average data access latency significantly.

Experiments confirm latency increases with physical distance.

Abstract

Erasure codes have been widely considered a promising solution to enhance data reliability at low storage costs. However, in modern geo-distributed storage systems, erasure codes may incur high data access latency as they require data retrieval from multiple remote storage nodes. This hinders the extensive application of erasure codes to data-intensive applications. This paper proposes novel caching schemes to achieve low latency in distributed coded storage systems. Experiments based on Amazon Simple Storage Service confirm the positive correlation between the latency and the physical distance of data retrieval. The average data access latency is used the performance metric to quantify the benefits of caching. Assuming that the future data popularity and network latency information is available, an offline caching scheme is proposed to find the optimal caching solution. Guided by the optimal scheme, an online caching scheme is proposed according to the measured data popularity and network latency information in real time. Experiment results demonstrate that the online scheme can approximate the optimal scheme well with dramatically reduced computation complexity.