Ken Claffey is the CEO of VDURA, brings in depth expertise in high-performance computing (HPC) and storage, having held govt roles at Seagate, Xyratex, Adaptec, and Eurologic. At Seagate, he led the Enterprise storage division, driving product innovation and market development. At Xyratex, he helped scale the corporate into a number one OEM supplier and constructed the ClusterStor HPC storage enterprise, which powered 40% of the world’s high supercomputers earlier than its acquisition by Cray/HPE. He additionally performed a key function within the sale of Xyratex to Seagate and has held management positions throughout product, operations, gross sales, and engineering.
As CEO of VDURA, Ken leverages his trade experience to drive innovation and development, positioning the corporate on the forefront of the evolving HPC and AI panorama.
VDURA is a knowledge storage and administration platform designed to help AI and high-performance computing (HPC) workloads. Using a shared-nothing scale-out structure, it goals to mix excessive efficiency with knowledge sturdiness at scale. The platform is adaptable for deployment in on-premises, cloud, and hybrid environments, providing storage options for data-intensive functions whereas emphasizing reliability and ease of use for enterprise operation.
You’ve had a outstanding profession spanning management roles at Seagate, Xyratex, Adaptec, and Eurologic. What are some pivotal moments that formed your method to innovation within the knowledge storage trade?
Essentially the most pivotal second of my profession was beginning the HPC enterprise at Xyratex, the place we created the world’s first HPC storage equipment optimized from the uncooked system stage to each layer of the software program stack. This included the {hardware} platform, the working system, the native file system, the parallel file system, networking, and even the consumer software program. That have formed my perception within the significance of deeply optimizing each layer to maximise infrastructure effectivity. Reaching this requires not solely a stable technical understanding of every layer, but in addition a grasp of the underlying enterprise drivers. All through my profession, I’ve been constructing and refining this distinctive mixture of technical and enterprise insights, which continues to tell my method to innovation within the trade.
What impressed the transition from Panasas to VDURA, and the way does the rebranding mirror the corporate’s present imaginative and prescient and objectives?
The transition from Panasas to VDURA displays a pure evolution of our firm’s mission to satisfy the altering wants of AI and HPC functions. Whereas we’re happy with Panasas’ legacy as a pioneer in parallel file techniques, it grew to become clear that the way forward for knowledge infrastructure required a brand new method—one which emphasizes flexibility, sturdiness, and ease. VDURA represents this subsequent chapter. The title itself—combining “velocity” and “sturdiness”—encapsulates our deal with delivering quick, dependable, and cost-effective storage options which are essential to trendy AI-driven workloads. Our rebranding additionally marks a strategic shift to a software-first mannequin, permitting VDURA to supply broader {hardware} selections and extra tailor-made options for our clients. By decoupling our software program from proprietary {hardware}, we’re creating new alternatives for purchasers to realize the very best efficiency and resilience throughout any deployment.
Are you able to share the story behind the creation of the VDURA Information Platform and the important thing challenges you aimed to deal with within the AI and HPC panorama?
The VDURA Information Platform was developed to deal with a essential and rising hole in knowledge infrastructure for AI and HPC workloads: balancing excessive efficiency with enterprise-grade sturdiness. As AI-driven functions proceed to evolve, knowledge platforms should deal with large volumes of various knowledge whereas making certain reliability, scalability, and seamless entry. Our platform builds on the inspiration of Panasas’ parallel NAS know-how, incorporating important developments corresponding to a totally parallel, microservices-based structure and our proprietary VeLO (Velocity Layered Operations) and VPOD (Virtualized Protected Object Machine) applied sciences. These improvements ship unmatched effectivity and sturdiness throughout hybrid, cloud-native, and on-premises environments, supporting various workloads from GPU-intensive AI coaching to billions of small file operations. What units VDURA aside is our capability to unify a number of storage media inside a single structure, offering the flexibleness to stability value and efficiency primarily based on buyer wants. The platform’s emphasis on scalability, sturdiness, and ease ensures organizations can meet the more and more demanding necessities of contemporary AI and HPC functions whereas optimizing their knowledge infrastructure investments.
What does “sturdiness” imply within the context of contemporary knowledge storage, and why is it more and more essential for AI-driven workloads?
Sturdiness in trendy knowledge storage ensures that knowledge stays intact, uncorrupted, and retrievable even after sudden occasions like outages or {hardware} failures. For AI-driven workloads, sturdiness is essential because of frequent checkpointing, large knowledge writing, and the reliance on correct datasets for mannequin coaching. VDURA prioritizes sturdiness via multi-layered knowledge safety, together with erasure coding and hybrid storage architectures that stability efficiency and sturdiness. This ensures that organizations can preserve knowledge integrity whereas scaling their infrastructure. With out sturdiness, even the quickest techniques threat knowledge loss, which might compromise insights and result in pricey disruptions.
The VDURA Information Platform emphasizes scalability and ease for AI and HPC functions. How do you see these components reshaping the way forward for knowledge infrastructure?
As AI and HPC workloads develop extra complicated, organizations want scalable options that scale back operational overhead with out compromising efficiency or reliability. VDURA’s Information Platform emphasizes simplicity by providing a unified structure that eliminates the necessity to handle disparate techniques. Scalability is equally essential. VDURA permits organizations to tailor their infrastructure to particular workload wants by using a multi-media storage method—leveraging flash storage for high-speed efficiency and HDD for cost-effective capability. This flexibility permits companies to stability efficiency and value whereas sustaining sturdiness and knowledge integrity. By offering intuitive and adaptable options, VDURA empowers organizations to focus extra time and sources on advancing their AI and HPC initiatives, relatively than managing storage complexity.
How does VDURA’s integration with options just like the Atlas AI Cluster improve the capabilities of organizations seeking to scale their AI operations?
VDURA’s integration with the Atlas AI Cluster simplifies and accelerates the deployment of AI factories by offering a totally optimized storage resolution tailor-made to the wants of AI workloads. The Atlas AI Cluster incorporates high-bandwidth, low-latency networking, NVIDIA’s Grace Hopper Superchips, and VDURA’s AI-optimized Information Platform, which provides superior metadata dealing with and enter/output operations per second (IOPS) for sooner knowledge entry. For us at VDURA, this partnership displays our dedication to constructing future-ready knowledge infrastructure that may deal with the large knowledge calls for of AI functions. By working carefully with companions like SourceCode and Eviden, we’re making certain that organizations can deploy cutting-edge AI infrastructure sooner and with higher confidence.
What are the most important challenges organizations face in adopting AI and HPC applied sciences, and the way is VDURA serving to to beat them?
Adopting AI and HPC applied sciences presents a number of challenges round scaling infrastructure, managing prices, and making certain knowledge sturdiness throughout their environments. Many organizations battle to stability efficiency with sturdiness as their workloads develop. VDURA helps handle these points by providing versatile, scalable storage options that scale back operational bills whereas making certain knowledge integrity for essential functions.
What developments do you foresee shaping the information storage and HPC markets within the subsequent decade?
Within the subsequent decade, knowledge sturdiness will turn into a key focus for organizations, as the price of knowledge loss or corruption takes heart stage in evaluating infrastructure investments. Companies will shift their priorities from availability to making sure knowledge integrity and persistence over time, recognizing the monetary dangers tied to unreliable techniques. VDURA may be very well-positioned to satisfy this demand with sturdy, scalable storage options that assist organizations keep away from the pricey penalties of downtime and knowledge loss.
With VDURA supporting industries like life sciences, power, and federal functions, how do you tailor your options to satisfy the distinctive calls for of those sectors?
VDURA tailors its options to the distinctive calls for of those industries by specializing in flexibility, sturdiness, and efficiency. These sectors generate large volumes of essential knowledge that require each high-speed entry and long-term integrity. Our hybrid storage structure permits organizations to optimize for each efficiency and capability—utilizing flash storage for real-time knowledge processing and HDD for cost-effective, sturdy storage. Our options additionally handle key regulatory and operational challenges in these industries by making certain knowledge stays safe, obtainable, and protected against loss or corruption. As our software program is decoupled from {hardware}, we give organizations the flexibleness to decide on the infrastructure that most closely fits their workload, enabling them to scale and adapt as their wants evolve.
Thanks for the nice interview, readers who want to study extra ought to go to VDURA.