A cluster is a combination of computers that are networked and run specific software for load balancing, redundancy, and high availability. Clustering in a distributed system emphasizes the scalability and fault tolerance of the system. Each node in the cluster performs a subset of tasks, and the cluster as a whole provides services to the outside world. If one node fails, other nodes can take over its work, ensuring the overall stability and continuity of the system.
Oracle recently announced the world's first Zettascale cloud cluster, a milestone event that represents a significant increase in cloud computing service capabilities. The cloud cluster, called Zettascale, plans to integrate up to 131,072 NVIDIA Blackwell chips, which will greatly improve data processing speed and efficiency. This technology upgrade is expected to significantly improve Oracle's competitiveness in the cloud computing market and could have a positive impact on the share prices of related hardware vendors.
It is reported that Oracle's new supercomputer cluster can be configured with Nvidia's Hopper or Blackwell GPUs for AI and HPC, as well as different networking devices. "As enterprises, researchers, and countries race to innovate with AI, they seek access to powerful computing clusters and AI software," said Ian Buck, vice president of hyperscale and high-performance computing at Nvidia. NVIDIA's full-stack AI compute platform on Oracle's widely distributed cloud will deliver AI computing power at an unprecedented scale to drive global AI efforts and help organizations around the world accelerate research, development, and deployment. ”
Figure: Oracle unveils a new type of cluster that brings a lot of performance to AI (Source: Oracle).
With the launch of the Zettascale cluster, Oracle can now deliver peak performance of up to 2.4 ZettaFLOPS, made possible by its Cloud Infrastructure (OCI). This level of performance is unprecedented, enabling Oracle's cloud computing services to handle the largest AI workloads. In addition, the cluster supports a wide range of GPUs, including NVIDIA H100, H200, B200, and GB200, as well as NVIDIA's fifth-generation NVLink technology and NVLink Switch, all for high-speed communication between GPUs.
Oracle's move not only demonstrates its ambitions in the field of high-performance computing (HPC), but also provides a robust AI infrastructure for businesses and researchers around the world. This is a huge boon for industries that require large-scale data processing and complex computing tasks, such as healthcare, scientific research, and financial services. Through this high-performance cloud computing service, enterprises can train and deploy AI models more flexibly and efficiently while maintaining data sovereignty.
It is worth noting that Oracle's announcement comes against the backdrop of fierce competition in the cloud computing market, with other major cloud service providers such as AWS, Azure, and Google Cloud constantly improving their service capabilities. With the launch of this Zettascale cluster, Oracle is undoubtedly taking a leading position in the field of high-performance computing in cloud computing.
Industry insiders say the upcoming OCI Super Cluster will use Nvidia's GB200 NVL72 liquid-cooled cabinet, which has 72 GPUs. According to Oracle, Nvidia's Blackewell GPUs will be available in the first half of 2025, but it's unclear when OCI will offer fully loaded Blackwell driver clusters.
Overall, the launch of Oracle's Zettascale cloud cluster is not only a demonstration of the company's technical prowess, but also a strong signal for the future development of the entire cloud computing industry. With the continuous improvement of cloud computing service capabilities, we have reason to expect more innovative applications and services to emerge to further promote the digital transformation of all walks of life.
