Understanding Data Storage and Ingestion for Large-Scale Deep Recommendation Model Training

TL;DR

This paper analyzes Meta's large-scale data storage and ingestion pipeline for deep learning training, highlighting bottlenecks, resource usage, and opportunities for hardware and infrastructure improvements.

Contribution

It provides a detailed characterization of Meta's DSI pipeline at scale, including infrastructure design, resource demands, and bottleneck analysis, with insights for optimization.

Findings

Identified hardware bottlenecks in DSI systems

Demonstrated the resource intensity of data preprocessing during training

Highlighted the need for heterogeneous DSI hardware and improved scheduling

Abstract

Datacenter-scale AI training clusters consisting of thousands of domain-specific accelerators (DSA) are used to train increasingly-complex deep learning models. These clusters rely on a data storage and ingestion (DSI) pipeline, responsible for storing exabytes of training data and serving it at tens of terabytes per second. As DSAs continue to push training efficiency and throughput, the DSI pipeline is becoming the dominating factor that constrains the overall training performance and capacity. Innovations that improve the efficiency and performance of DSI systems and hardware are urgent, demanding a deep understanding of DSI characteristics and infrastructure at scale. This paper presents Meta's end-to-end DSI pipeline, composed of a central data warehouse built on distributed storage and a Data PreProcessing Service that scales to eliminate data stalls. We characterize how hundreds of models are collaboratively trained across geo-distributed datacenters via diverse and continuous training jobs. These training jobs read and heavily filter massive and evolving datasets, resulting in popular features and samples used across training jobs. We measure the intense network, memory, and compute resources required by each training job to preprocess samples during training. Finally, we synthesize key takeaways based on our production infrastructure characterization. These include identifying hardware bottlenecks, discussing opportunities for heterogeneous DSI hardware, motivating research in datacenter scheduling and benchmark datasets, and assimilating lessons learned in optimizing DSI infrastructure.