23 The advent of Artificial Intelligence (AI), and now Machine Learning (ML) and Deep Learning (DL), is driving amazing innovation and breakthroughs across many industries. AI is often compared to a rocket ship, with learning algorithms as its engine and data as jet fuel. At its core, ML/DL depends on training and inference, which involve feeding huge amount of (training) data to the algorithm so it can adapt and improve, in order to learn how to make decisions (inferences) on his own. The algorithms perform better and become more accurate as the training data sets grow. This is why AI/ML/DL workloads require data at such a massive scale. Simply put: the more data, the better outcome (as measured by accuracy, speed, etc). Storage Bottlenecks The ability to aggregate and orchestrate massive amount of data surrounding AI/ML/DL workloads is critical. Data sets frequently reach multi-petabyte scale, with performance demands that could saturate the whole infrastructure. When dealing with such large-scale training and test data sets, addressing storage bottlenecks (latency and/or throughput issues) as well as capacity limitations/barriers are key success factors. AI/ML/DL workloads require a storage architecture that can keep data flowing through the pipeline, with both stellar raw I/O performance and capacity scaling prowess. The storage infrastructure needs to keep up with exceedingly demanding requirements across all stages of the AI/ML/DL pipeline. The adequate solution is a storage infrastructure specifically built for raw speed and unlimited scale. AI Data Nodes AI Data Node from HPE is an integrated, packaged solution tailor-made for AI/ML/DL workloads. It offers the best of both worlds, combining HPE Scalable Object Storage with Scality RING and WekaIO Matrix™ parallel file System, running side-by-side on HPE Apollo 4200 Gen10 server. In a nutshell, AI Data Node from HPE implements a two-tier storage architecture, featuring a high-performance flash tier paired with a scalable object storage tier, all in-place and under a single namespace. By combining the inherent storage performance of HPE Apollo 4200 Gen10 server, the raw performance of WekaIO parallel file system, and the unlimited scale and efficiency of Scality RING, the AI Data Node fulfills the raw storage performance and storage capacity scaling demands imposed by AI/ML/DL workloads, and delivers the data to fuel and accelerate your AI rocket ship. To learn more, we invite you to download the technical white paper detailing the Reference Architecture for AI Data Node from HPE.
