VAST Data is providing its AI OS, including all-flash storage, to the SK Telecom Petasus AI Cloud which provides GPU-as-a-Service (GPUaaS) in South Korea’s sovereign AI cloud.
SK Telecom, South Korea’s main telecommunications supplier, is evolving into an AI company. It is building an AI infrastructure based on a Haein cluster of Supermicro GPU servers with Nvidia Blackwell GPUs, networking and storage. A Petasus GPUaaS AI cloud will use this cluster. It virtualizes the GPU, networking, and storage cluster, and provides it to multiple tenants.
VAST Data is contributing to the storage part of the Petasus AI Cloud.
DK Lee, VP and Head of SK Telecom’s AI DC Lab, stated: “VAST Data’s unified architecture has been instrumental in helping us move from legacy bare-metal deployments to a fully virtualized, production-grade AI cloud.” He said the VAST AI OS provides performance, simplicity, flexibility, and fast scalability.
“With VAST, we’re enabling a GPUaaS platform that meets the exacting needs of government, research, and enterprise AI customers in South Korea.”
Sunil Chavan, VP APAC at VAST Data added his view: “From our earliest conversations, it was clear that SKT needed cutting-edge infrastructure to match the speed and complexity of enterprise-grade uptime and nation-state inference and training. By eliminating traditional bottlenecks around data movement, provisioning, and security, VAST is enabling SKT to launch a sovereign and secure AI infrastructure that offers speed and flexibility at scale for Korea.”
The Petasus cloud will enable AI model development and deployment within South Korea’s borders. Users – tenants – need not provision their own bare metal infrastructure. They specify the GPU resources they need and a virtual GPU server system, a virtual cluster, is spun up for them in as little as 10 minutes. It automatically provisions and isolates GPU and storage resources, including their networking fabrics, to match each tenant’s specific requirements. Each virtual cluster will, it’s claimed, match bare metal performance.
An SK Telecom diagram illustrates the cloud components and multi-tenancy;
This GPUaaS cloud includes VAST’s disaggregated, shared-everything (DASE) platform on Supermicro servers built with NVIDIA’s HGX architecture.
Tenant workloads will be Isolated with data privacy and performance guarantees. The system provides multi-protocol data access with no need for client-side gateways or proprietary protocols. SK Telecom says the system will have carrier-grade reliability.
SK Telecom’s Haein Cluster has been selected for the Ministry of Science and ICT’s “AI Computing Resource Utilization Enhancement(GPU Rental Support) Program” in Korea, and will be used for the development of national AI foundation models.
Read more in a VAST blog: SK Telecom and VAST Data: Breaking the GPU Virtualization Barrier, and also in a VAST podcast.
Comment
DDN has been deeply involved with SK Telecom – witness this Youtube video – and SK Telecom execs have presented at DDN events about the development of the SK Telecom AI Cloud, its Petasus virtualization, and DDN storage use. Whether DDN is still involved with the SK Telecom AI Cloud remains to be seen.
