Japan's leading container data center provider accelerates innovation with the integration of KAYTUS' next-generation AI servers.

The client extends its data center with KAYTUS 6U2S AI servers, streamlining maintenance and accelerating business growth.

Customer Introduction

The client is a leading containerized data center operator in Japan, specializing in providing cloud computing support to local customers. Currently, the company is operating two data centers in Tokyo and Niigata Prefecture, with the Niigata facility renowned for its robust power infrastructure, supporting over 500 servers. In recent years, the client has experienced rapid growth, becoming one of the largest container data center providers in Japan. As a consequence of this rapid expansion, the existing server clusters are no longer sufficient to meet the increasing demand.

Project Background

Driven by the ongoing digital transformation, the Japanese data center industry is experiencing a new phase of rapid expansion. Containerized data centers, valued for their efficiency, flexibility, and scalability, have become the key solution for many enterprises and cloud service providers. As a leading operator of containerized data centers in Japan, the client has encountered challenges in upgrading its hardware infrastructure to keep pace with this growth. Particularly with the increasing adoption of generative AI and large language model (LLM) applications, the demand for high-performance computing has massively increased. To address these increasing computational demands, the client has adopted the next-generation hardware solution, which ensures continued competitiveness and sustainable growth of their data centers.

Challenges

1. Existing setup is insufficient to support the increasing business demands.

Due to the rapid advancement of large language models (LLMs), the client identified that their existing hardware infrastructure was insufficient to support the increasing demand for high computational power. Hereafter, the client made the decision to upgrade the hardware in their data center, with a particular focus on enhancing computing capacity for LLM training and the development of generative AI applications. To address the growing need for AI computing power among their container data center customers, the client ultimately chose to establish a new cluster, considering crucial factors such as capacity, stability, and manageability.

2. Ensure prompt delivery
Since the existing computing capacity at the customer data center is not sufficient and the requirement of rigorous computing demand is increasing at a rapid pace, therefore the customer needed to deploy high-performance AI servers to run the new project within three months to meet the growing demand of existing and future potential customers. However, the industry average delivery time is about ten months which is not feasible for this project.

KAYTUS Solution

After extensively analyzing the challenges of customers' current data centers and aligning them with the actual business demands, KAYTUS proposed a cluster solution built with the next-generation KAYTUS 6U2S flagship servers.  First, multiple rounds of internal testing were conducted based on the client’s specific requirements to ensure the solution met performance and stability expectations. Secondly, to guarantee timely delivery, materials were prepared in advance, with both the required parts and spare parts readied simultaneously to allow immediate production after testing.   Following a comprehensive evaluation of several server suppliers, the client determined that the KAYTUS 6U2S AI-powered cluster solution delivered superior performance and stability for high-density computing tasks, while also meeting the special requirement for rapid delivery, solving the problem of explosive user demand growth.

The KAYTUS V2 server, the latest generation from KAYTUS, successfully adapted to Intel's new Sapphire Rapids CPUs, overcoming numerous technical challenges in the process. This innovation fully met the client's stringent requirements for new hardware and delivered the necessary cloud computing power. The KAYTUS 6U2S server, specifically engineered for high-performance computing, offers the following key advantages:

1. Strong scalability
Supports expansion up to 8 GPU cards, catering to diverse computing demands and fully meeting the client's extensive   high-performance storage requirements.

2. Exceptional stability and dependability
Equipped with redundant power supplies and an advanced cooling design, to ensure stability during long-term high-load operations and fulfilling the client's requirement for continuous, reliable performance.

3. Superior energy efficiency
Employs advanced power management technology to significantly reduce energy consumption, aligning with green data center standards, and meeting the client's low-power consumption requirements, thus substantially reducing electricity costs.

Customer Benefits

After the deployment of a container data center composed of KAYTUS 6U2S flagship AI servers, computing power increased by 2.8 folds, while operational efficiency more than doubled, significantly enhancing the client's data processing capabilities. In terms of LLM training and generative development, computing speed nearly tripled.  Due to the high energy efficiency design of the new equipment, the overall energy consumption of the data center was reduced by about 20%, leading to substantial reductions in operational costs. The boost in computing power and stability also resulted in a notable rise in customer satisfaction, with new customers expected to increase by more than 50% in 2024.