“The new server from Dell Technologies, 'Dell PowerEdge,' supports 67% enhanced performance compared to existing servers by integrating artificial intelligence (AI) workloads.”
Chris Kelly, senior vice president of Dell Technologies APJC (Asia Pacific, Japan, and China), introduced the new server 'Dell PowerEdge,' which is part of the 'AI Ready Data Center' (a data center designed to efficiently support AI technologies), during an online media briefing on the 8th. Kelly noted, “By using the Dell PowerEdge server to consolidate existing legacy servers, we can reduce power consumption by more than 50% compared to the previous systems and significantly improve server space efficiency.”
Dell Technologies also unveiled a new data storage system called 'Dell PowerStore' following the Dell PowerEdge server launch. Dell PowerStore is an enterprise-grade storage solution offered by Dell Technologies that focuses on scalability and automation, designed to help corporations process and manage large-scale data.
Notably, a distributed architecture has been applied to the data storage system. Kelly explained, “If data is operated solely on a single operating system (OS), there can be disadvantages to efficiently utilizing resources. It is advantageous in terms of efficiency to handle only the necessary data in a distributed manner without operating the entire system,” adding, “In an era of AI where data is surging, we introduced a distributed architecture in the data storage system to prevent network saturation and utilize the data efficiently.”
Kelly noted that graphics processing units (GPUs) play a crucial role in the implementation of distributed architecture. He stated, “Since distributed architecture is a method where multiple computer systems or servers collaborate to process data, GPUs are utilized as critical resources for processing large data sets and performing complex calculations in a distributed manner.” He added, “While a sudden increase in data processing in a single architecture has the drawback of rapidly consuming GPUs, the advantage of distributed architecture is that GPU usage is spread out.”
Kelly also expressed his views on water-cooled (a method that uses water to cool) data center cooling. He said, “Water cooling is (efficient but) still difficult to generalize,” suggesting that “air cooling seems to be (still) appropriate.”