Designing Storage Infrastructures for AI Workloads

Presented by

Storage Switzerland and Panasas

About this talk

Artificial Intelligence (AI) and Machine Learning (ML) workloads are fundamentally different from any other workload. These workloads deal in data sets measured in dozens of petabytes of capacity, and large percentages of it can go from dormant to active at any moment. The high storage requirement and highly active data set make using all-flash arrays or even large flash caches impractical. IT planners need to rethink their designs to create infrastructures that are highly parallel and leverage high capacity hard disk drives to strike the right balance of performance and capacity. These infrastructures also need metadata efficiency, since AI and ML workloads are metadata heavy. Join us to learn: * Why all organizations need to prepare for AI and ML * The challenges of AI/ML at scale * Why current storage infrastructures can’t meet the AI/ML at scale challenge * Requirements of an At-Scale AI/ML Storage Infrastructure

Related topics:

More from this channel

Upcoming talks (1)
On-demand talks (413)
Subscribers (49358)
StorONE's mission is to make Storage Heroes by providing IT professionals with a solution that meets all the objectives of software-defined storage, plus adds data protection and resiliency that is 10X better than any other technology on the market. At the same time, it increases storage system performance 10X over industry standard. The result is StorONE's Enterprise Storage Platform. StorONE founders are proven, successful entrepreneurs who boast over 50 patent awards. Top-tier industry luminaries and strategic investors back the company. The S1 Storage Platform changes the perception of data center storage from an IT cost center to a resource that provides organizations with a critical competitive advantage. The StorONE headquarters are in New York, with offices in Fort Worth TX, Tel Aviv, and Singapore.