Architectural Support for Efficient Data Movement in Disaggregated Systems

TL;DR

This paper presents DaeMon, a software-transparent mechanism that reduces data movement overheads in disaggregated systems by enhancing hardware with specialized engines and employing adaptive data migration strategies, improving performance and cost-efficiency.

Contribution

It introduces DaeMon, the first robust, software-transparent system for efficient data movement in disaggregated architectures, combining hardware enhancements and adaptive migration techniques.

Findings

Significantly improves system performance over page-granularity data movement.

Reduces data access costs in disaggregated systems.

Enhances scalability and robustness across various network conditions.

Abstract

Resource disaggregation offers a cost effective solution to resource scaling, utilization, and failure-handling in data centers by physically separating hardware devices in a server. Servers are architected as pools of processor, memory, and storage devices, organized as independent failure-isolated components interconnected by a high-bandwidth network. A critical challenge, however, is the high performance penalty of accessing data from a remote memory module over the network. Addressing this challenge is difficult as disaggregated systems have high runtime variability in network latencies/bandwidth, and page migration can significantly delay critical path cache line accesses in other pages. This paper introduces DaeMon, the first software-transparent and robust mechanism to significantly alleviate data movement overheads in fully disaggregated systems. First, to enable scalability to multiple hardware components in the system, we enhance each compute and memory unit with specialized engines that transparently handle data migrations. Second, to achieve high performance and provide robustness across various network, architecture and application characteristics, we implement a synergistic approach of bandwidth partitioning, link compression, decoupled data movement of multiple granularities, and adaptive granularity selection in data movements. We evaluate DaeMon in a wide variety of workloads at different network and architecture configurations using a state-of-the-art accurate simulator and demonstrate that DaeMon significantly improves system performance and data access costs over the widely-adopted approach of moving data at page granularity.