AI Development, Operations, and Multi-Cloud Heterogeneous Management No Longer a Concern: INFINITIX Optimizes Enterprise Resource Efficiency

Reprinted from DigiTimes

Wei Yuning / Taipei News 2022-07-19

Cloud adoption is seen as a necessary strategy for enterprises to drive digital transformation, with features such as high flexibility, on-demand functionality, and optimized resource allocation. However, due to the diversity of IT technologies, enterprises may adopt hybrid cloud platforms, virtual machines, or containers for different business needs. In the past, there was no user-friendly and simple unified management platform for this heterogeneous multi-cloud architecture, resulting in inefficient management. INFINITIX's solution can help operators enhance management efficiency.

INFINITIX is a pioneer in multi-cloud and hybrid cloud management platforms. General Manager Chen Wenyu pointed out that in the early stages of cloud platform commercialization, they designed the CloudFusion multi-cloud/hybrid cloud management platform. In 2017, the company collaborated with the National Center for High-Performance Computing to develop the GPU Cloud-TWGC platform, which is the predecessor of the supercomputer "Taiwan Cypress II". Subsequently, INFINITIX further optimized cloud and AI management technologies and launched the "AI-Stack" machine learning development and operations management platform to help enterprises lower the deployment threshold of new-generation heterogeneous cloud IT architectures and improve management efficiency.

The introduction of the AI-Stack platform is mainly to solve the IT management pain points encountered by enterprises in this era. Chen Wenyu stated that the industrial environment is changing rapidly, and IT architecture needs to keep pace with the times. In the past, most enterprise IT systems were mainly on-premise. Later, as cloud platforms gradually became popular, hybrid clouds combining public and private elements gradually became the mainstream of enterprise IT architecture. This multi-cloud heterogeneous IT architecture has created various management pain points. In on-premise systems, virtual machines and containers are now mostly used in combination. When there were only virtual machines in the past, IT system management was relatively simple. With the addition of containers, due to their multiple usage states and the need to adjust and manage related functions and parameters within each container, optimizing the performance of on-premise IT systems has become more complex. For cloud systems, due to the diverse charging methods of public cloud functions, different departments may choose different services. The combination of functional service fees and accompanying storage and computing resource fees often result in very complex billing items, making it difficult for managers to quickly and accurately grasp the cost structure of IT systems and optimize the use of enterprise resources.

To address these issues, INFINITIX's CloudFusion multi-cloud/hybrid cloud management platform can integrate mainstream public clouds in the market through a universal interface. Enterprises can self-manage virtual machines and object storage through a single interface, and quickly schedule networks for load balancing. For on-premise systems, INFINITIX starts with GPU management. Chen Wenyu pointed out that GPUs have become the mainstream computing architecture for enterprises developing AI and edge computing. The AI-Stack platform makes GPU management easier and more convenient. During the AI model training stages such as data collection and labeling, AI-Stack can clearly understand the computing needs of each link and perform optimal scheduling. With the assistance of this platform, not only can GPU usage efficiency be increased from the usual 30% to 90%, but enterprises can also add GPUs on a project-by-project basis due to the transparency of computing resource data, solving the problem of excessive idle computing resources in the past.

To optimize GPU management performance, INFINITIX actively cooperates with GPU industry leader NVIDIA. Now, as INFINITIX joins the NVIDIA Partner Network (NPN) to become Taiwan's first NVIDIA solution advisor, channel partner Zero One Technology is also promoting value-added cooperation between the AI-Stack platform and NVIDIA, creating a GPU resource-sharing management platform for enterprise AI developers that combines flexibility and efficiency, optimizing cross-departmental AI work efficiency.

INFINITIX currently has multiple successful cases. For example, a financial institution shortened its AI project timeline by completing the development, testing, and deployment stages in a single interface. Additionally, many manufacturers and scientific research units have achieved good results after implementing CloudFusion and AI-Stack. Chen Wenyu stated that with the popularization of cloud and AI, hybrid cloud platforms and heterogeneous computing architectures will rapidly land in various types of enterprises in the future. The resulting system complexity will become an IT management pain point for enterprises. In the future, INFINITIX will strengthen platform functions in response to market demands and continue to play a solid role in supporting enterprise AI development and cloud management.

INFINITIX focuses on solving challenges brought by virtualization, containerization, microservices, edge computing, hybrid cloud management, heterogeneous IT environments, and artificial intelligence for enterprises. It integrates heterogeneous cloud management with popular open-source AI deep learning frameworks and development tool environments to provide a one-stop AI machine learning cloud platform solution, AI-Stack.