Project Overview

This project is located at Jomo Kenyatta University of Agriculture and Technology (JKUAT) in Kenya. JKUAT is Kenya’s top-ranked public comprehensive university focusing on engineering and technology and an academic benchmark in the field of engineering and ICT in East Africa. It has in-depth layouts in engineering technology, information science and AI applications, with a high degree of internationalization, and has established long-term cooperative relationships with universities from China, Germany, the United Kingdom and other countries, as well as global technology enterprises.

In recent years, with the full advancement of the “Digital Teaching + Scientific Research Computing” strategy, the university has rapidly built an online learning system, campus all-in-one card system, and laboratory data management platform, and plans to deploy a small-scale AI research cluster to support teaching and research in data science and machine learning. However, the university’s original computer room infrastructure has seriously lagged behind and can no longer match the rapid growth of business needs, becoming a core bottleneck restricting digital transformation.

GOTTOGPOWER designed and delivered a solution deploying multiple sets of GT-DC8000 and GT-DC5000 micro modular data center systems tailored to JKUAT’s actual needs and the local environmental characteristics in Africa. The solution integrates online double-conversion UPS systems and high-precision computer room precision air conditioners, realizing one-stop delivery of the entire link from power distribution and cooling to monitoring and IT hosting. The project has now completed installation and commissioning

Core Customer Pain Points

In-depth research in the early stage of the project found that the university was not facing a simple problem of insufficient computer room capacity, but a concentrated outbreak of structural infrastructure contradictions common in the digital transformation of African universities.

First, power quality issues. The power grid infrastructure in the suburbs of Nairobi is weak, with large voltage fluctuations, instantaneous voltage sags and short-term power outages occurring frequently. This has led to frequent restarts of core servers, and multiple instances of data inconsistency or even loss in the educational administration system and online examination system, seriously affecting normal teaching order.

Second, thermal management out of control. The original computer room used scattered ordinary commercial air conditioners for cooling, resulting in chaotic airflow organization and severe mixing of hot and cold air. It could not cope with the thermal load of high-density servers, leading to obvious local hot spots. The inlet air temperature of some cabinets was long higher than the standard range, and the equipment failure rate increased significantly.

Third, insufficient scalability. The university’s traditional computer room adopted a fixed construction model without modular design. With the launch of the AI laboratory and online course platform, the number of IT equipment increased by 65% within 6 months. Each expansion required shutdown construction, affecting teaching business for an average of 2-3 days each time.

Fourth, weak operation and maintenance capabilities. The university’s IT team is small in scale and lacks professional data center operation and maintenance experience. Relying on manual inspection, it cannot realize real-time monitoring and predictive maintenance, and the average time for fault discovery and handling exceeds 4 hours.

Solution Architecture Design

In response to the above problems, GOTTOGPOWER adopted an overall architecture of “modular distributed micro data center + hierarchical hosting” instead of the traditional single computer room expansion model.

The GT-DC8000 data center serves as the core data center module, deployed in the main data center area of the campus, used to host the university’s core business systems, including the educational administration management platform, database system, storage system and AI training computing nodes, providing the highest level of power supply and cooling guarantees.

The GT-DC5000 data center serves as an extended edge module, deployed in the laboratory building and the School of Computer Science area, used to support teaching experiment platforms and local computing resources, realizing nearby business access and load distribution, and reducing the pressure on the main computer room.

This architecture fundamentally solves the three core problems of centralized management, distributed hosting and future scalability, perfectly adapting to the characteristics of wide business distribution and rapid growth in universities.

In terms of power supply system design, all micro modules uniformly adopt online double-conversion UPS systems, realizing complete electrical isolation between input mains power and output loads, completely eliminating the impact of voltage fluctuations, harmonic interference and instantaneous power outages on IT equipment. The UPS system adopts a modular redundant design, with 30% redundant capacity reserved in the initial configuration, ensuring uninterrupted system operation when a single module fails, and supporting online hot-swappable expansion without shutdown.

In terms of cooling system, high-precision computer room dedicated precision air conditioners are used, optimizing the airflow path through overhead air return, enabling cold air to accurately cover the server air inlet surfaces. The system has precise constant temperature and humidity control capabilities, avoiding efficiency losses caused by hot and cold air mixing.

At the same time, the system stabilizes the computer room environment within the range recommended, effectively reducing server hardware failure rates and data error risks, and maximizing the service life of equipment.

Key Technical Implementation Details

During the project implementation, GOTTOGPOWER focused on solving three key engineering-level difficulties.

First, power supply reliability verification under extreme power environments. In view of the frequent power outages in the Nairobi power grid, multiple rounds of full-load power outage switching tests were conducted on site. The test results showed that the system can achieve true 0-millisecond uninterrupted power supply, and all servers and business systems did not experience any restart or interruption when the mains power was completely cut off.

Second, cooling redundancy capability verification under high temperature environments. Considering the large temperature difference between day and night in Nairobi and high ambient temperatures in some seasons, the system underwent a continuous 72-hour thermal load test under full-load operation conditions. The temperature fluctuation in the core area of the computer room was controlled within ±1.5℃, and no local overheating occurred.

Third, zero-impact rapid deployment implementation. All GT-DC8000 data centers and GT-DC5000 data centers modules are factory prefabricated and pre-tested before leaving the factory. On-site work only includes module assembly, busbar connection, refrigerant pipeline connection and system commissioning, compressing the traditional 3-6 months construction period to 45 days, and the entire process did not affect the normal operation of the university’s existing business systems.

Dynamic Environment Monitoring and Operation & Maintenance Optimization

Considering the limited operation and maintenance capabilities of the customer’s IT team, this project focused on enhancing the usability and intelligence level of the Data Center Infrastructure Management system.

The system conducts unified centralized monitoring of UPS input and output status, battery health, load rate, air conditioner operating parameters, computer room temperature and humidity distribution, access control and fire protection systems, and provides an intuitive visual management interface. When an abnormal situation occurs, the system can automatically trigger hierarchical alarms and push them to administrators in real time via email, SMS and mobile APP.

This is particularly critical in the African education industry, as the response speed of on-site engineers is limited, and remote visual operation and maintenance capabilities directly determine system reliability. Through this system, the university has realized the transformation of operation and maintenance mode from “passive repair” to “active early warning”, reducing the average fault response time from the original 4 hours to less than 15 minutes, greatly reducing operation and maintenance difficulties and labor costs.

Project Achievements and Customer Value Enhancement

After the project delivery, the university realized a structural upgrade from “distributed computer rooms” to “unified micro modular data centers”.

• In terms of operation and maintenance, the dynamic environment system has enabled the university to have complete data center visual management capabilities for the first time, greatly reducing reliance on manual inspection, realizing 7×24-hour unattended operation of the data center, and reducing operation and maintenance labor costs by more than 60%.
• In terms of power supply reliability, key business systems have achieved zero unplanned interruptions throughout the year, completely solving the risk of teaching accidents caused by power problems.
• In terms of environmental stability, the computer room temperature control capability has been significantly improved, the average operating temperature of servers has decreased, the hardware alarm rate has dropped by more than 40%, and the expected service life of equipment has been extended by more than 30%.
• In terms of scalability, the module supports on-demand expansion, which can be expanded online without affecting existing businesses, meeting the growth needs of future AI courses and research projects.

Project Conclusion

This project is not only a simple equipment delivery, but also an infrastructure architecture innovation targeting the pain points of digital transformation in the African education industry. Through the combined application of GT-DC8000 and GT-DC5000 micro modular data centers, GOTTOGPOWER has provided African universities with a replicable standardized integrated solution of “high-reliability power supply + precise environmental control + modular expansion + intelligent operation and maintenance”.

This case fully verifies the unique advantages of modular data centers in the context of weak power infrastructure, shortage of professional talents and rapid growth of digital demand in Africa. It has strong replicability and industry promotion value, and is also applicable to various application scenarios such as hospitals, government agencies and small and medium-sized cloud computing nodes.