Panasas

Founded in 2000, Panasas, which rebranded to VDURA in May 2024, is a long-standing entity in the high-performance computing (HPC) storage sector. The company was established by Dr. Garth Gibson, a prominent computer scientist from Carnegie Mellon University who was instrumental in developing the RAID taxonomy for data storage. Gibson's background in storage systems, including his work on RAID and at the Parallel Data Laboratory at Carnegie Mellon, directly informed the company's focus. Panasas was created to commercialize concepts around object-based storage and parallel file systems to address the growing demands of data-intensive applications.

VDURA operates in the AI and HPC data infrastructure market, providing software-defined storage solutions to clients in academic research, government, energy, life sciences, and manufacturing. Its business model centers on selling its proprietary software, PanFS, bundled with commodity hardware, and has recently introduced a software-as-a-service (SaaS) model to offer more flexibility. The company generates revenue through these sales and has established a robust partner channel, which accounted for over 50% of revenue in the past year. Public records show the company has received over $149 million in funding across multiple rounds, with investors including Centennial Ventures, Mohr Davidow Ventures, and others.

The core of VDURA's offering is the PanFS® parallel file system, an operating environment designed to manage massive data volumes for AI and HPC workloads. PanFS integrates a parallel file system, volume manager, and a scalable RAID engine, orchestrating multiple storage servers to function as a single entity. This architecture allows for the parallel processing of file components, overcoming the bottlenecks of traditional network-attached storage (NAS) systems. Key features include Dynamic Data Acceleration, which adapts to mixed-workload environments, and the DirectFlow® protocol, which enables compute clients to communicate directly and in parallel with the storage nodes for higher throughput. The system is designed for linear scalability, meaning performance and capacity increase proportionally as more nodes are added. To ensure data integrity, PanFS employs per-file RAID 6+ with triple parity, offering advanced data protection. The platform supports hybrid storage configurations of SSD and HDD, and recent updates include S3 compatibility for cloud workflow integration and support for dual-actuator hard drives to boost performance without shifting entirely to more expensive all-flash storage. Keywords: high-performance computing, parallel file system, data storage, software-defined storage, AI infrastructure, object storage, scale-out NAS, unstructured data, data management, enterprise storage