Harnessing the power of data-centric servers for the AI-driven future
Around the world, data centers have undergone a remarkable transformation to keep pace with the ever-growing demands of a thriving digital landscape. From the core to the cloud and the edge, the modern enterprise data center is more likely than not a hybrid, multicloud environment that spans multiple locations.
Despite the dramatic changes in architecture, the foundation of data centers continues to rely on powerful servers with cutting-edge technology, optimized to meet the growing computational demands of businesses.
A data-centric view of workloads
The traditional approach to server design has typically revolved around a general-purpose machine, with flexible configurations and upgrade options to meet varying workloads. This worked well to support everyday business processes or solutions with relatively stable requirements, such as enterprise resource planning or customer relationship management systems.
However, the past few years have seen the emergence of more data-centric workloads in the form of data simulation, data modeling and artificial intelligence (AI). In addition, the volume, variety and velocity of data that enterprises are dealing with have surged, while demands to utilize data have increased exponentially.
Despite growing recognition of the importance of data, just a small percentage of the enormous exabytes of data created every day is computed with analytics or AI, according to Chris Kelly, senior vice president for data center solutions, Asia Pacific and Japan (APJ), at Dell Technologies. This can be attributed to the lack of processing capabilities within many enterprises.
“As a business continues to evolve and modernize to stay competitive, its requirement around speed and agility—getting to outcomes faster—is highly dependent on its ability to generate and act on intelligence,” Kelly says.
“Data plays a significant role in training an AI model for any organization,” he adds. “And as businesses turn to AI, it has amplified the focus on making sure they have the computational infrastructure to turn data into insights sooner. Data-centricity is really about valuing data or making investments into the systems that give you the ability to turn data into insight and get faster value from data.”
To leverage the growth of data to their advantage, the onus is on enterprises to ensure they have the scalable infrastructure to get the most out of their data. Achieving this calls for data-centric servers equipped with faster memory and I/O bandwidth—as well as the security features to handle today’s complex workloads.
Purpose-built for AI
Unlike generic servers designed for atypical usage scenarios, data-centric servers are built from the ground up to be highly scalable platforms engineered with the latest technologies across microprocessors, memory, networking, storage and accelerators. This must be incorporated into every aspect of design, manufacturing, delivery, deployment and operation, allowing customers to take advantage of AI regardless of whether it is deployed in the core, in the cloud or at the edge.
Designing such a server starts with ensuring high throughput for greater I/O and faster operation of data. It also entails workload profiling of traditional applications and emerging techniques in AI, data analytics and data science use cases, ensuring that the resultant platform is capable of meeting existing applications and future requirements.
Moreover, the movement of data is considered as it moves around within the computing platform: as it is loaded from storage to memory, processed on the CPU, and then returns back to the memory and long-term storage—or across the network. The result is an advanced server that offers faster processing of data, optimized AI and machine learning capabilities, and the ability to meet the performance criteria of businesses without creating bottlenecks.
A partner for sustainability
“Delivering more than 300 world records to date, the exceptional performance of AMD EPYC processors is enabling companies to make use of their ever-increasing volume of data to help transform their businesses,” says Peter Chambers, managing director of sales for APJ at AMD. “This level of performance when combined with cutting-edge security and outstanding energy efficiency helps to deliver customers a leap forward in total cost of ownership and sustainability for these next-gen data-centric workloads.”
There is no question that the next wave of digital transformation will be driven by next-gen AI and data-centric workloads. However, these workloads consume greater amounts of energy. Enterprises recognize this and are prioritizing sustainability in their strategic planning.
One key step is to align corporate sustainability goals with technology infrastructure investments and adopt energy-efficient solutions that balance performance, cost and environmental impact. This can be achieved by partnering with a vendor that takes environmental sustainability seriously—such as Dell Technologies.
“The new 4th Generation AMD EPYC processors provide 50% more core density with up to 47% better performance per watt over the previous generation, enabling a highly efficient data center that helps you reduce your business’s carbon footprint,“ says Kelly.
“Improved thermal design helps to be able to improve the energy efficiency of the platform. And this comes through innovations such as a T-shaped motherboard for optimum airflow through the compute platform and an updated bezel to improve airflow from the front to the rear of the server,” he adds.
“We make every effort to reduce waste and reuse available resources in manufacturing, and we work very closely with our supply chain partners to address sustainability throughout their entire ecosystem, both in packaging and delivery.”
Dell Technologies has also built an extensive AI Innovation Lab dedicated to working with customers and helping them understand best practices and assessments. The knowledge gained is then used to inform the design of future Dell PowerEdge solutions, helping customers stay on the leading edge of new and emerging use cases.
Designed to deliver accelerated business results while lowering costs, power and server footprint, next-generation Dell PowerEdge servers powered by 4th Generation AMD EPYC processors push workload boundaries with tailored IT and business solutions to help organizations keep pace with the proliferation of AI and data workloads.
With up to 96 cores in a single socket, the new 4th Generation AMD EPYC processors enable customers to deploy fewer efficient and more powerful servers to continue to meet their computing needs. This offers greater flexibility within the data center that can be leveraged to address business sustainability goals and drive real-world dividends.
Start with your requirements
To ascertain their future requirements, enterprises should start with current workloads when considering their data-centric journey. This is done by analyzing the existing environment and collecting information about current server deployments for a deeper understanding.
On this front, Dell Technologies offers a free, agentless tool that lets enterprises stream workload data from their server infrastructure to an Analytical Engine, which then measures and analyzes the workload characteristics of the existing infrastructure. By establishing a baseline and extrapolating from it to incorporate AI and future workloads, enterprises are better empowered to determine what a future investment might look like.
This baseline also gives enterprises a meaningful starting point to understand how they can potentially deploy AI within existing applications or leverage AI to build a more immersive experience that improves their business operations and allows them to respond more effectively to customers.
On this front, Kelly recommends that APJ businesses sign up for an “Art of the Possible” workload discovery workshop at one of the Dell Solution Centers in the region (Singapore, Shanghai, Sydney or Tokyo).
“The workshop helps enterprises understand the potential benefit from successful adoption of emerging workloads and where others are seeing real value, be it industry bodies, a peer in their industry or potential collaborators in the industry,” Kelly says.
The workshop is designed to help guide enterprises on best practices, sizing and proof of concept for server deployments.
