Supermicro Introduces a Rack Scale Total Solution for AI Storage to Accelerate Data Pipelines for High-Performance AI Training and Inference
- Supermicro launches a full stack optimized storage solution for AI and ML data pipelines
- The solution supports petascale capacities and high-performance data delivery
- President and CEO Charles Liang claims the solution can deliver 270 GB/s of read throughput and 3.9 million IOPS per storage cluster
- The solution aims to maximize AI time-to-value by keeping GPU data pipelines fully saturated
- The solution delivers multi-petabyte data for AIOps and MLOps in production environments
- None.
Insights
The launch of Supermicro's new full stack optimized storage solution marks a significant advancement in the field of data management for AI and ML applications. This development is poised to have a substantial effect on the competitive landscape of the data storage market, particularly for companies that require large-scale, high-performance storage solutions to keep up with the demands of AI and ML workloads.
From a market perspective, this innovation by Supermicro could potentially lead to an increase in the company's market share within the data storage sector. By offering a solution that addresses the specific needs of AI and ML, Supermicro is positioning itself as a leader in a niche but rapidly growing segment. The ability to handle petascale capacities and deliver high read throughput and IOPS is critical for organizations looking to scale their AI operations and Supermicro's product appears to meet these requirements effectively.
Furthermore, the integration with NVIDIA's HGX H100 GPUs and the use of PCIe 5.0 and E3.S storage devices suggest a strategic partnership that could enhance Supermicro's offering. This collaboration might also positively influence NVIDIA's positioning in the AI infrastructure market. As more companies invest in AI and ML, the demand for such high-performance solutions is expected to grow, which could benefit both Supermicro and NVIDIA.
Supermicro's announcement of its new storage solution tailored for AI training and inference workloads could have notable financial implications for the company. The focus on reducing total cost of ownership (TCO) and increasing AI performance addresses a critical pain point for businesses, potentially leading to a strong value proposition and higher sales volumes.
Analyzed from a financial standpoint, the ability to deliver a scalable solution that ranges from a minimum deployment to hundreds of petabytes suggests that Supermicro is targeting not just large enterprises but also potentially tapping into the mid-market segment. This scalability could result in a diverse customer base and a steady revenue stream. The emphasis on reducing implementation risks and enabling faster model training could also lead to customer retention and long-term contracts, which are favorable for predictable and stable cash flows.
Investors should monitor Supermicro's financial performance in subsequent quarters to assess the impact of this product launch on revenue growth and profit margins. It's also important to consider the capital expenditure associated with developing and marketing such a high-tech solution, as well as the operational costs of supporting a potentially complex product ecosystem.
The technical aspects of Supermicro's new storage solution demonstrate a significant leap in the data storage technology used for AI and ML workloads. The solution's ability to deliver 270 GB/s of read throughput and 3.9 million IOPS per storage cluster is a testament to its high-performance capabilities. For context, Input/Output Operations Per Second (IOPS) is a performance measurement used to benchmark the speed at which storage devices can read and write data. The figures mentioned are indicative of a system that can handle intense data workloads, which is essential for AI applications that require rapid data processing and analysis.
The use of PCIe 5.0 technology is another critical aspect to highlight, as it represents the latest standard in high-speed data transfer interfaces, offering double the bandwidth of PCIe 4.0. This could significantly reduce data bottlenecks and improve overall system efficiency. Additionally, the mention of E3.S storage devices points to a form factor that is optimized for density and performance in enterprise storage systems. The collaboration with WEKA Data Platform software also indicates a comprehensive solution that is likely to be well-received by industries that depend on high-speed data analytics.
Turn-Key Data Storage Solution for Large Scale AI Training and Inference – Hundreds of Petabytes in a Multi-tier Solution Supports the Massive Data Capacity Required and High-Performance Data Bandwidth Necessary for Scalable AI Workloads
"With 20 PB per rack of high-performance flash storage driving four application-optimized NVIDIA HGX H100 8-GPU based air-cooled servers or eight NVIDIA HGX H100 8-GPU based liquid-cooled servers, customers can accelerate their AI and ML applications running at rack scale," said Charles Liang, president and CEO of Supermicro. "This solution can deliver 270 GB/s of read throughput and 3.9 million IOPS per storage cluster as a minimum deployment and can easily scale up to hundreds of petabytes. Using the latest Supermicro systems with PCIe 5.0 and E3.S storage devices and WEKA Data Platform software, users will see significant increases in the performance of AI applications with this field-tested rack scale solution. Our new storage solution for AI training enables customers to maximize the usage of our most advanced rack scale solutions of GPU servers, reducing their TCO and increasing AI performance."
For more information about the Supermicro Storage Solutions for AI, please visit https://www.supermicro.com/en/products/storage
Petabytes of unstructured data used in large-scale AI training processing must be available to the GPU servers with low latencies and high bandwidth to keep the GPUs productive. Supermicro's extensive portfolio of Intel and AMD based storage servers is a crucial element of the AI pipeline. These include the Supermicro Petascale All-Flash storage servers, which have a capacity of 983.04* TB per server of NVMe Gen 5 flash capacity and deliver up to 230 GB/s of read bandwidth and 30 million IOPS. This solution also includes the Supermicro SuperServer 90 drive bay storage servers for the capacity object tier. This complete and tested solution is available worldwide for customers in ML, GenAI, and other computationally complex workloads.
The new storage solution consists of:
- All-Flash tier - Supermicro Petascale Storage Servers
- Application tier – Supermicro 8U GPU Servers: AS -8125GS-TNHR and SYS-821GE-TNHR
- Object tier - Supermicro 90 drive bay 4U SuperStorage Server running Quantum ActiveScale object storage
- Software: WEKA Data Platform and Quantum ActiveScale object storage
- Switches: Supermicro InfiniBand and Ethernet Switches
"The high performance and large flash capacity of Supermicro's All-Flash Petascale Storage Servers perfectly complement WEKA's AI-native data platform software. Together, they provide the unparalleled speed, scale, and simplicity demanded by today's enterprise AI customers," said Jonathan Martin, president at WEKA.
Supermicro will present the optimized storage architecture in more detail in a webinar. To attend the webinar live on Feb. 1, 2024, or view the Supermicro and WEKA webinar on-demand, visit: https://www.brighttalk.com/webcast/17278/604378
Learn More about All Supermicro Storage Systems
*Raw value is based on vendor raw base capacity of 30.72TB. TB is base-10 decimal. Availability of 30.72TB E3.S SSD is subject to vendor availability.
About Super Micro Computer, Inc.
Supermicro (NASDAQ: SMCI) is a global leader in Application-Optimized Total IT Solutions. Founded and operating in
Supermicro, Server Building Block Solutions, and We Keep IT Green are trademarks and/or registered trademarks of Super Micro Computer, Inc.
All other brands, names, and trademarks are the property of their respective owners.
View original content to download multimedia:https://www.prnewswire.com/news-releases/supermicro-introduces-a-rack-scale-total-solution-for-ai-storage-to-accelerate-data-pipelines-for-high-performance-ai-training-and-inference-302045157.html
SOURCE Super Micro Computer, Inc.
FAQ
What company is launching the full stack optimized storage solution for AI and ML data pipelines?
What is the ticker symbol for Supermicro, Inc.?
What is the maximum read throughput the solution can deliver per storage cluster?
What is the maximum number of IOPS the solution can deliver per storage cluster?
What is the main aim of the solution?
Who claims that the solution can reduce TCO and increase AI performance?
