Adit Madan and Parviz Peiravi offer an overview of the Alluxio data orchestration layer that provides a unified data access layer for hybrid and multi cloud deployments, leveraging Intel® Optane™ Persistent Memory for higher performance caching at reduced cost. The data access layer enables distributed compute engines like Presto, TensorFlow, and PyTorch to transparently access data from various storage systems (including S3, HDFS, and Azure) while actively leveraging a multi-tier cache to accelerate data access.
A collaboration of Alibaba, Alluxio, and Nanjing University in tackling the problems of Deep Learning model training in the cloud. Our goal was to reduce the cost and complexity of data access for Deep Learning training in a hybrid environment, which resulted in over 40% reduction in training time and cost.
This article presents the collaborative work of Alibaba, Alluxio, and Nanjing University in tackling the problem of Artificial Intelligence and Deep Learning model training in the cloud. We adopted a hybrid solution with a data orchestration layer that connects private data centers to cloud platforms in a containerized environment. Various performance bottlenecks are analyzed with detailed optimizations of each component in the architecture.