The company's new approach to data center architectures aims to address the limitations of today's data platform. It uses shared persistent memory instead of slow storage for writes in combination with a single tier of all flash storage, bringing super-low latency for transactional writes and high throughput for analytical reads. This approach eliminates the historical tradeoff between latency and throughput. The company's vision is to unify files, objects, and tables as one unified namespace, blurring the distinction between data and metadata as they turn strings into things.
Key takeaways:
- VAST Data Inc. is developing a sixth data platform that can support intelligent applications in real time.
- The modern data stack will need to evolve to separate compute from data and enable compute to operate on a unified view of coherent and composable data elements.
- VAST Data's system uses shared persistent memory instead of slow storage for writes, combined with a single tier of all flash storage, to bring super-low latency for transactional writes and high throughput for analytical reads.
- The company's roadmap includes building out system- and application-level functionality, with the aim of unifying data types and metadata to provide access to AI-infused applications at scale.