Heterogeneous-Reliability Memory: Exploiting Application-Level Memory Error Tolerance

TL;DR

This paper introduces a methodology to measure application error tolerance, enabling tailored memory reliability in datacenters, which reduces costs while maintaining high server availability.

Contribution

It presents a new way to quantify application memory error tolerance, analyzes three workloads for error vulnerability, and proposes heterogeneous-reliability memory designs to optimize datacenter costs.

Findings

Memory error tolerance varies across applications.

Cost can be reduced by 4.7% with maintained 99.90% availability.

Heterogeneous-reliability memory designs are effective.

Abstract

This paper summarizes our work on characterizing application memory error vulnerability to optimize datacenter cost via Heterogeneous-Reliability Memory (HRM), which was published in DSN 2014, and examines the work's significance and future potential. Memory devices represent a key component of datacenter total cost of ownership (TCO), and techniques used to reduce errors that occur on these devices increase this cost. Existing approaches to providing reliability for memory devices pessimistically treat all data as equally vulnerable to memory errors. Our key insight is that there exists a diverse spectrum of tolerance to memory errors in new data-intensive applications, and that traditional one-size-fits-all memory reliability techniques are inefficient in terms of cost. This presents an opportunity to greatly reduce server hardware cost by provisioning the right amount of memory reliability for different applications. Toward this end, in our DSN 2014 paper, we make three main contributions to enable highly-reliable servers at low datacenter cost. First, we develop a new methodology to quantify the tolerance of applications to memory errors. Second, using our methodology, we perform a case study of three new data-intensive workloads (an interactive web search application, an in-memory key--value store, and a graph mining framework) to identify new insights into the nature of application memory error vulnerability. Third, based on our insights, we propose several new hardware/software heterogeneous-reliability memory system designs to lower datacenter cost while achieving high reliability and discuss their trade-offs. We show that our new techniques can reduce server hardware cost by 4.7% while achieving 99.90% single server availability.