NAMUTECH (CEO Chul Jung) announced on the 18th the release of version 4.8 of ‘Cocktail Cloud’, which features enhanced Graphics Processing Unit (GPU) capabilities.
Cocktail Cloud supports customized cloud construction environments tailored to the business environments of companies. It provides multi-cluster and hybrid cloud environments, allowing for the integrated management of all applications, physical servers, and cloud infrastructure through a single control point interface. The newly released version 4.8 has strengthened its GPU management features and backup and restore capabilities.
NAMUTECH’s Cocktail Cloud offers features such as GPU resource sharing and partitioning, monitoring, data backup and restoration of clusters, and fast and easy cluster provisioning through provisioning templates.
Through the GPU dashboard, administrators can monitor detailed GPU cluster usage, GPU utilization, GPU memory, temperature, power, and more. Additionally, they can view detailed GPU resource usage by node and the list of Pods utilizing GPU resources.
GPU sharing supports two methods: Time Slicing and Multi Instance GPU (MIG). Time Slicing allows for the physical GPU to be shared in a round-robin scheduler manner, enabling workloads to utilize GPU resources without waste maximally. Multi Instance GPU enables a single physical GPU to be partitioned into up to seven instances, increasing the efficiency of GPU resources.
The backup and restore feature allows for the storage of all resources of Cocktail Cloud and implements a safe backup and rapid recovery environment through scheduled automation of backups. Administrators can escape vendor lock-in by supporting a multi-cloud environment with various vendors such as Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP).
