Understanding data storage and ingestion for large-scale deep recommendation model training
Authors:
Top Cited Papers
Open Access
- 11 June 2022
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 1042-1057
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.Keywords
All Related Versions
This publication has 39 references indexed in Scilit:
- Production Machine Learning Pipelines2021
- Delta lakeProceedings of the VLDB Endowment, 2020
- CerebroProceedings of the VLDB Endowment, 2020
- Efficient User-Level Storage Disaggregation for Deep Learning2019
- The Datacenter as a Computer: Designing Warehouse-Scale Machines, Third EditionSynthesis Lectures on Computer Architecture, 2018
- The dataflow modelProceedings of the VLDB Endowment, 2015
- Leakage in data miningACM Transactions on Knowledge Discovery From Data, 2012
- DremelProceedings of the VLDB Endowment, 2010
- HiveProceedings of the VLDB Endowment, 2009
- ImageNet: A large-scale hierarchical image database2009