MaxMem: Colocation and Performance for Big Data Applications on Tiered Main Memory Servers

TL;DR

MaxMem is a user-space memory management system that optimizes performance and colocation for Big Data applications on tiered main memory systems by using fast memory miss ratio sampling.

Contribution

MaxMem introduces an application-agnostic, lightweight memory occupancy control mechanism based on memory miss ratios for tiered main memory management.

Findings

MaxMem achieves 11% higher throughput than HeMem.

MaxMem reduces latency by up to 80% compared to Linux AutoNUMA.

MaxMem improves performance in dynamic colocation scenarios.

Abstract

We present MaxMem, a tiered main memory management system that aims to maximize Big Data application colocation and performance. MaxMem uses an application-agnostic and lightweight memory occupancy control mechanism based on fast memory miss ratios to provide application QoS under increasing colocation. By relying on memory access sampling and binning to quickly identify per-process memory heat gradients, MaxMem maximizes performance for many applications sharing tiered main memory simultaneously. MaxMem is designed as a user-space memory manager to be easily modifiable and extensible, without complex kernel code development. On a system with tiered main memory consisting of DRAM and Intel Optane persistent memory modules, our evaluation confirms that MaxMem provides 11% and 38% better throughput and up to 80% and an order of magnitude lower 99th percentile latency than HeMem and Linux AutoNUMA, respectively, with a Big Data key-value store in dynamic colocation scenarios.