Advanced Computing in the Age of AI | Monday, October 3, 2022

New Apps Drive Demand for Compute Capacity 
Sponsored Content by Supermicro and Intel

The growing reliance on Big Data analytics, the expanded use of more sophisticated modeling, and the emergence of the Internet of Things as well as mainstream artificial intelligence applications is driving up compute requirements in most organizations.

Faced with data centers that are tight on space and power, indiscriminately adding compute capacity will not work. What’s needed is a move to servers that deliver more compute power, while taking up less rack space and using less electricity for power and cooling.

Challenges in supporting today’s enterprise applications

New applications are significantly increasing compute requirements placing even more demands on data centers. There are estimates that in the near-term, data centers will have to support up to 300 percent increased workloads.

As a result, IT infrastructures are being stressed to the max thanks to new enterprise applications.

While moving the applications to the cloud is an option, most critical workloads are still run on-premises. One annual industry study found that the percentage of workloads running in enterprise-owned and operated data centers has stayed at a stable 65 percent since 2014.

This means most organizations will need to add capacity in their data centers. In a survey or IT leaders asking about their plans for this year, 27 percent said their organizations plan to increase build out or support of IT infrastructure to support new business opportunities. Another 30 percent cited increased workforce demands as the driver for a bigger focus on infrastructure.

Selecting the right solutions and the right technology partner

When trying to meet the compute requirements of modern enterprise apps, the biggest constraints are the space and power limitations in today’s data centers.

Certainly, new servers based on the latest generation processors offer a way to increase raw compute capacity. However, in most organizations, the right server is one that delivers the highest performance per watt.

This is an area where Supermicro can help.

Supermicro offers the IT industry’s broadest portfolio of new generation servers, demonstrating the highest performance, energy efficiency, and cost-effectiveness.

The company has been a leader in delivering systems for the most demanding applications. In 2013, Supermicro introduced Hadoop-optimized 1U and 2U Twin and 4U FatTwin™ servers. These server solutions used Intel’s then-new Xeon Phi coprocessors to cost-effectively accelerate the performance of many high-end applications.

The servers are based on Supermicro's patented Twin Architecture. These high-performance, high-density systems feature optimum airflow for energy efficient cooling, easy maintenance and high availability with hot-swappable nodes and redundant power supply modules. These systems also feature the latest CPU and interconnect technologies for maximum processing performance and data throughput.

The current X11 generation Twin family of servers are based on Intel Xeon Scalable processors. The line includes:

  • The Supermicro TwinPro, which provides exceptional throughput, storage, networking, I/O, memory and processing capabilities in a 2U form factor. The TwinPro doubles the performance of a traditional 2U system and houses four nodes to reduce data center footprint and includes redundant Titanium Level Digital (96%) Power Supplies for maximum power efficiency.
  • The Supermicro BigTwin™, the first and only 2U multi-node system supporting the highest performance processor, up to 3TB of memory, all-flash NVMe storage, and an incredible 30% better thermal capacity over comparably outfitted single-node systems. The 5th generation of Supermicro's Twin architecture is fully optimized for today's diverse workloads and architected to support future innovations in data center technology.
  • The new generation X11 Supermicro FatTwin™, a high-density 8/4 hot-plug node SuperServer® system available with a variety of memory capacities, HDD technologies, M.2 SSD NVMe, PCIe alternatives, and networking capabilities. The Supermicro FatTwin represents a revolution in Green Computing and is highly efficient by design. It is designed to support critical applications while reducing data center TCO. It extends the compute and storage capabilities of Supermicro's existing Twin SuperServer® systems to achieve increased performance and power efficiency.

To further address performance issues, Supermicro offers NVMe (Non-Volatile Memory Express) support. NVMe, as a logical device interface, was designed to capitalize on the low latency and internal parallelism of flash-based storage devices such as SSDs.

The primary benefits of Supermicro server solutions with NVMe support include improved performance (throughput and latency), hot-swap capability, ability to support many true hot-swap NVMe drives per server, and cost-effectiveness compared to add-on card-based solutions. Supermicro’s hot-swap support for NVMe SSD drives allows easy addition of storage capacity through the addition of SSDs, the replacement of existing SSDs with higher capacity units, or replacement of failed drives, while the server system is in operation. This allows a company to easily grow its storage capacity, while minimizing server down time. Both capabilities are essential for today’s critical enterprise applications.

In summary, the Supermicro Twin family of servers can help organizations run their most critical applications while making the most of their existing data center footprint. New data centers also benefit from Twin server architecture, as they are highly energy-efficient and easy to install and manage. This results in a low TCO while freeing up IT staff to work on more critical tasks.

To learn more about how to get the most out of your data center space and meet the growing compute requirements of today’s enterprise applications, visit:

Add a Comment