Case Study: Siemens Healthineers achieves high-performance, fault-tolerant AI storage with Quobyte

Siemens Healthineers scales AI storage for 2 billion medical images with Quobyte

Siemens Healthineers, a global leader in medical technology, faced a significant challenge scaling the storage infrastructure for its AI supercomputer, Sherlock. Their existing systems could not efficiently handle the massive, multi-terabyte datasets required for daily deep learning experiments without time-consuming data copies to scratch space. They needed a high-performance, fault-tolerant, and production-ready parallel file system that could run on their existing commodity hardware. After testing other solutions, they found Quobyte met these requirements.

By implementing the Quobyte parallel file system, Siemens Healthineers created a scalable, unified storage solution that eliminated the need for scratch space, allowing AI workloads to access data directly at high speed. The results included seamless scalability, simplified data governance with secure tenant domains, and comprehensive fault-tolerance with multi-cluster replication. Quobyte enabled the training of massive AI models on over half-a-billion medical images and ensured business continuity for over 150 critical R&D projects annually.

Siemens Healthineers