Data-centric Organizations need both Performance and Flexible Data Management

Converged AI and HPC have arrived in 2020 and commercial IT organizations are putting data at the core of their business model. Just in the HPC market alone, Intersect360 found that around 90% of HPC organizations were either running machine learning (56%) or investigating/planning its introduction. Gartner claims that 60% of organizations are adapting their business model to AI and a 2019 ESG. survey showed that 49% of IT organizations state that “data is their business,with another 31% expecting to offer datacentric products in the next 2 years.Yet, how to reliably accomplish that goal often remains elusive. The same Gartner survey reported a 50% failure rate for these projects. Performance, as ever, is important and DDN’s EXAScaler systems continue to outperform forthcoming competitor offerings with over 63 Million IOPs per Rack with DDN’s AI400X and over 140GB/s per rack in HDD-only sequential performance. But delivering this performance potential to today’s workloads is the real target. Datacentric organizations are investing heavily in more powerful compute platforms, often relying on GPUs. Keeping systems operating at 100% utilization is not an easy proposition without at-scale experience and concerted engineering effort in optimizing for containerized workloads, AI frameworks, GPU platforms, extreme-scale CPU and fast networks. Underutilized infrastructure is a recipe for failure in a world where competitive stakes are high, investments are large and the amount of data to be analyzed is vast. A data management vendor must deliver performance no matter how large the requirement nor how diverse the applications.