November 11, 2024

PEAK:AIO Tackles GPU Data Bottlenecks with High-Performance Storage Solutions for AI Workloads at SC24

Written by

As organizations increasingly invest in GPU infrastructure for AI and HPC workloads, the challenge of data throughput bottlenecks has become more pressing than ever.

In this exclusive VMblog interview, Roger Cummings, CEO at PEAK:AIO, discusses how his company is transforming commodity hardware into ultra-fast AI data servers, enabling organizations to maximize their GPU investments without the burden of expensive legacy storage systems.

Ahead of SC24, Cummings reveals how PEAK:AIO's software-defined approach delivers market-leading speed from a single node while prioritizing cost-effectiveness and sustainability - a crucial consideration as the industry grapples with growing GPU demands in terms of both performance and power consumption.

VMblog:  If you were giving an SC24 attendee a quick overview of the company, what would you say? How would you describe the company?

Roger Cummings:  PEAK:AIO empowers organizations to fully leverage their GPU investments by delivering high-performance storage specifically designed for AI and HPC workloads. PEAK:AIO turn commodity hardware into ultra-fast AI data servers, offering a cost-effective solution that removes data throughput bottlenecks. This enables faster and more efficient innovation, helping our clients achieve new levels of productivity and performance.

PEAK:AIO enables organizations to maximize their GPU investments, prioritizing AI innovation and outcomes instead of wasting resources on outdated storage systems. Our software transforms standard servers into incredibly fast data solutions, bypassing the expenses and complexities of traditional storage. This leads to superior performance and faster results, allowing more funds to be directed towards projects and innovation rather than being squandered on legacy infrastructure.

VMblog:  Your company is sponsoring participating in event in this year's SC24. Can you talk about what that sponsorship your participation looks like?

Cummings:  At SC24, the PEAK:AIO team is actively engaging in a range of activities, including speaking engagements, live demonstrations, press events, and collaborative exhibit support with our partners. We are thrilled to showcase our solution, designed to address real-world AI and HPC workload challenges. Our focus is on delivering industry-leading performance, scalability, and cost-effectiveness for our clients to reach new heights of innovation and efficiency.

At SC24, we are excited to collaborate with leading vendors like Western Digital, Solidigm, and Hammerspace. Each vendor excels in different markets, however our technology enhances them, whether by boosting performance, lowering costs, simplifying use or significantly cutting power consumption. 

VMblog:  What kind of message will an attendee hear from you this year? What will they take back to help sell their management team and decision makers? Explain your technology.

Cummings:  PEAK:AIO strongly believe that the increasing GPU demands in terms of performance and power cannot be met by the existing trend of simply increasing storage resource, a traditional way of scaling performance.  These new demands by growing GPU usage requires not simply a new solution, but new thinking, and at PEAK:AIO we are committed to delivering new levels of performance but within realistic packages.  Our focus on developing technology to create more efficient ultra-performance solution within more usable and manageable ‘blocks' is at the forefront of our design focus.

VMblog:  Thinking about your company's technologies, what are the types of problems you solve for an SC24 attendee?

Cummings:  We tackle essential data bottlenecks in AI and HPC workloads, especially in GPU-intensive environments where data throughput is critical to productivity and innovation. Yet, space and power constraints are a challenge. Our technology addresses these challenges without relying on costly and complex multi-node solutions, enabling organizations to efficiently manage large-scale data demands and enhance performance. 

VMblog:  While thinking about your company's technology and solution offerings, can you give readers a few examples of how your offerings are unique? What are your differentiators? What sets you apart from the competition?

Cummings:  PEAK:AIO's software is uniquely designed to:

(1) Transform commodity hardware into a high-performance storage server optimized specifically for AI workloads.

(2) Unlike other solutions, we provide the flexibility to scale performance from a single node, delivering market-leading speed without proprietary hardware lock-in. This flexibility, combined with cost-efficiency, truly differentiates us and empowers organizations to achieve exceptional performance without sacrificing budget or adaptability. 

(3) Couple this with the 2025 vision of supplying over 6PBs within a single 2U, we are enabling projects from standalone medical applications to hyperscaler-level to implement realistic solutions.

 

VMblog:  How can attendees of the event find you? What do you have planned at your booth this year? What type of things will attendees be able to do at your booth?

 

Cummings:  Attendees can find us at our partners' booths, including Solidigm, Western Digital, Hammerspace and Boston Limited. We'll be showcasing our AI Data Server technology, with our team available to answer questions and conduct in-depth technical discussions.

 

VMblog:  Have you sponsored SC24 in the past? If so, what is it about this show that keeps you coming back as a sponsor?

Cummings:  This isn't our first time at SC24, and we return because it's a premier event for connecting with thought leaders and innovators in the HPC and AI space. SC24 offers a unique platform for us to showcase our advancements, gather valuable feedback, and build relationships with the people shaping the future of high-performance computing.

VMblog:  Do you have any speaking sessions during the event? If so, can you give us the details?

Cummings:  Yes, we have a few sessions with our partners, where we'll discuss our ability to drive storage performance from a single node to extensive scale-out environments. Together, we deliver the performance and scalability that customers demand, with a focus on cost-effectiveness and environmental sustainability.

VMblog:  What major industry challenges or technological bottlenecks do you see your solutions addressing in the high-performance computing (HPC) landscape, particularly in relation to emerging AI/ML workloads?

Cummings:  Our solution tackles the data throughput challenges in GPU-driven environments. As AI and ML workloads advance, they demand faster, more dependable data access to match the computational speed of GPUs. This need is increasingly pushing closer to the edge. PEAK:AIO has a strong track record of helping customers overcome these challenges, enabling GPUs to operate at their full potential and ensuring high productivity for AI/ML applications.

VMblog:  With the increasing focus on sustainable computing, how is your organization innovating to deliver better performance per watt, and what new energy-efficient technologies will you be showcasing at SC24?

Cummings:  PEAK:AIO's software-defined approach allows organizations to build high-capacity, densely configured nodes, significantly reducing the overall infrastructure required. By minimizing the number of nodes, simplifying network complexity, and saving physical space, our solution contributes to a more sustainable computing environment. This results in better performance per watt and supports greener, more efficient data centers.

VMblog:  The convergence of HPC and AI continues to reshape the industry. How are your solutions enabling organizations to bridge traditional HPC workloads with new AI/ML demands, and what unique capabilities will you be demonstrating at the event?

Cummings:  PEAK:AIO bridges the gap between HPC and AI by providing storage solutions robust enough for traditional HPC workloads while optimized for the unique demands of AI. Our platform offers the flexibility to manage diverse, data-intensive tasks, supporting everything from health and scientific research to complex AI model training-all on a single, high-performance infrastructure.

VMblog:  As we see growing demand for "AI supercomputers," how is your company adapting its product strategy to address this evolving market segment? What differentiates your approach from others in the space?

Cummings:  We're adapting to the demand for 'AI supercomputers' by focusing on scalable, software-defined storage that enhances the flexibility of these high-performance systems. Our solution enables organizations to build high-density nodes on any commodity hardware, avoiding vendor lock-in. This makes it easier to scale and adapt as AI demands evolve, setting us apart from competitors who rely on proprietary solutions.

VMblog:  What role do you see quantum computing playing in the future of supercomputing, and how is your organization preparing for this potential paradigm shift?

Cummings:  While quantum computing shows great potential, today's AI and HPC workloads continue to depend on GPU and CPU architectures. PEAK:AIO remains committed to optimizing data movement and storage for these current technologies, while also exploring how our flexible, software-driven architecture could seamlessly integrate with quantum systems as they become more widely available.

VMblog:  With the rise of cloud-native HPC solutions, how is your company helping customers navigate the balance between on-premises, cloud, and hybrid deployments? What new tools or services will you be introducing at SC24 to address these needs?

Cummings:  PEAK:AIO enables hybrid deployments with flexible, software-defined storage that can be implemented both on-premises and in cloud environments. This approach allows organizations to enjoy consistent high performance, whether leveraging local hardware or scaling out in the cloud. At SC24, we'll showcase how our solution seamlessly supports diverse deployment needs, helping customers navigate the balance between on-premises, cloud, and hybrid setups.

VMblog:  Data movement and storage continue to be critical challenges in HPC. What innovations is your company bringing to market to address these bottlenecks, and what performance improvements can customers expect?

Cummings:  PEAK:AIO's software is designed to deliver ultra-fast data access and movement, even with a single-node configuration. Our latest innovations include support for QLC staging, ensuring long-term data availability in high-performance environments. PEAK:AIO's intelligent staging capability allows customers to build the most efficient and cost-effective infrastructure tailored to their business needs, helping them overcome data movement and storage bottlenecks with significant performance gains.

VMblog:  How is your organization contributing to democratizing access to HPC resources, particularly for smaller organizations and research institutions? What initiatives or solutions will you be highlighting at SC24 in this area?

Cummings:  PEAK:AIO is committed to democratizing access to high-performance computing by providing a cost-effective, high-performance solution that supports early-stage AI innovation. Our technology lowers the entry barrier to AI and HPC, empowering research institutions and startups to compete with industry leaders without requiring substantial infrastructure investments. At SC24, we'll be showcasing how our solutions enable broader access to advanced computing resources.

VMblog:  As security concerns continue to grow in the HPC space, especially with the integration of AI workloads, what approaches and technologies is your company developing to ensure both performance and protection?

Cummings:  Security is deeply integrated into the PEAK:AIO's solution, extending to our RAID configurations. As AI and HPC workloads continue to evolve, the importance of security and governance grows. Earlier this year, PEAK:AIO introduced PEAK:ARCHIVE, a solution that enhances audit and security capabilities to safeguard AI data and model integrity. This ensures accountability and protection as these critical workloads are executed.

David Marshall

David Marshall has been involved in the technology industry for over 19 years, and he's been working with virtualization software since 1999. He was able to become an industry expert in virtualization by becoming a pioneer in that field - one of the few people in the industry allowed to work with Alpha stage server virtualization software from industry leaders: VMware (ESX Server), Connectix and Microsoft (Virtual Server).

Through the years, he has invented, marketed and helped launch a number of successful virtualization software companies and products. David holds a BS degree in Finance, an Information Technology Certification, and a number of vendor certifications from Microsoft, CompTia and others. He's also co-authored two published books: "VMware ESX Essentials in the Virtual Data Center" and "Advanced Server Virtualization: VMware and Microsoft Platforms in the Virtual Data Center" and acted as technical editor for two popular Virtualization "For Dummies" books. With his remaining spare time, David founded and operates one of the oldest independent virtualization news blogs, VMblog.com. And co-founded CloudCow.com, a publication dedicated to Cloud Computing. Starting in 2009 and continuing all the way to 2016, David has been honored with the vExpert distinction by VMware for his virtualization evangelism.

Gold Sponsors

Graid Technology


Hammerspace


Komprise


PEAK:AIO


StoreONE

Latest Videos