Innogrid Completes AI Big Data Analytics Infrastructure Build for the Korea SMEs and Startups Institute
IT DAILY ·
✦ AI Summary
Innogrid announced the completion of the Korea SMEs and Startups Institute's 'GPU-based AI big data analytics infrastructure build' project.
The project focused on advancing the analytics environment of the 'Small and Medium Enterprise Big Data Platform (SIMS),' and Openstackit and TabCloudit were applied.
It converted the existing closed on-premises and physical PC-centered structure to a high-performance GPU environment, and also established server and storage integration, GPU virtual machines, and a monitoring framework.
Innogrid announced on the 1st that it had completed the Korea SMEs and Startups Institute's 'GPU-based AI big data analytics infrastructure build' project.
The project was pursued with a focus on advancing the analytics environment of the 'Small and Medium Enterprise Big Data Platform (SIMS).' The solutions applied were the private cloud platform 'Openstackit' and the cloud management platform 'TabCloudit.'
The existing structure was centered on a closed on-premises environment and physical PCs, and through this project it was converted to a high-performance GPU environment for developing advanced analytics models. In the process, it also met the requirements for introducing domestic virtualization software that had obtained a security function verification certificate from the National Intelligence Service.
To that end, equipment was installed in the systems room of the Korea SMEs and Startups Institute, and a network including fiber optic cables was configured. As a result, an infrastructure environment was established in which cloud servers and storage are organically linked.
At the platform layer, the private cloud platform and the cloud management platform were applied at the same time, securing both a virtualization-based infrastructure and a management framework. For operating functions, snapshots and replication were applied, and dynamic scaling was used to support stable operations and future scalability.
At the computing layer, GPU virtual machines (VM) equipped with GPUs were configured for use in AI big data analytics research. It also supported the function of assigning multiple GPUs to a single virtual machine and the configuration of high-performance development VMs, making it possible to allocate resources according to the nature and scale of each project.
At the management layer, a system was set up to check host and virtual server resource status in real time. A monitoring system was also established to predict scaling timing based on usage trends, supporting decisions on when to expand.
Based on this build experience, Innogrid is proposing migration directions to customers considering a shift from foreign virtualization environments. The migration plan is based on establishing a framework that takes into account the characteristics of existing VM workloads and the importance of services, while the expansion structure is aimed at gradually extending from private cloud to multi- and hybrid-cloud environments and also to GPU-based AI workloads.
Innogrid CEO Kim Myeong-jin said that when converting an analytics environment to the cloud, it is important to design the system while reflecting both existing workloads and new GPU resources. Kim said the project was an example of a cloud architecture transition across the entire analytics environment, including server and storage integration, GPU virtual machine configuration, and a monitoring framework.
In September, Innogrid launched Openstackit 3.5, its private cloud solution. Openstackit 3.5 reflects an advancement in the company's own virtualization technology.
Source: IT DAILY · Kwon Young-seok
Original: https://www.itdaily.kr/news/articleView.html?idxno=241954
References
This article was produced with the help of an automated content generation algorithm.
Source: IT DAILY
View originalThis article was summarized and organized by BizCrush based on the original article from IT DAILY. For exact quotations and full details, please refer to the original article.