As AI becomes an indispensable part of organizational innovation and competitive advantage, the demand for efficient and scalable infrastructure is more urgent than ever before. The collaboration between NVIDIA and DDN Storage is setting new standards in this field. By integrating NVIDIA BlueField data processors into DDN EXAScaler and DDN Infinia and using them in innovative ways, DDN Storage is transforming data centric workloads.
The synergy between NVIDIA’s advanced data processing and DDN’s powerful storage solutions can optimize AI workflows, enhance operations, improve resource utilization, and support multi tenancy in AI environments.
In this article, we will focus on introducing the integration advantages of Infinia。
Integrated DPU storage solution
DDN Infinia is a software defined data platform that utilizes the powerful capabilities of BlueField-3 DPU to effectively manage data centric workloads, particularly in accelerating computation and generative artificial intelligence. This integration enhances multi tenancy, improves operational efficiency, and strengthens data protection. This makes it an ideal solution for organizations to use artificial intelligence and cloud technology to drive innovation and operational agility.
Their solution includes several key components:
Uninstalling data processing
Accelerate storage performance
Improve efficiency
Support multi tenancy
Enhance security
Enhanced Expansion
Uninstall data processing
BlueField DPU reduces the burden on the CPU by taking over data processing tasks, thereby freeing up computing resources and improving overall system performance. This storage and security task offloading can improve CPU usage efficiency, reduce latency, and accelerate data processing speed。
Accelerate storage performance
The storage solution of DDN is supported by BlueField DPU, which can enhance the storage performance of AI workloads. With the advanced data processing capabilities of BlueField DPU, these solutions achieve higher throughput and system response speed to accelerate AI applications.
NVIDIA GPUDirect Storage (GDS) helps establish a direct data path between GPU platforms and storage, thereby minimizing system memory traffic, increasing bandwidth, and reducing CPU load to optimize AI workflows.
increase of efficiency
Traditional storage systems perform flash management on general-purpose x86 CPUs RAID、 Various tasks such as access control and encryption. However, as network speed and security requirements increase, the efficiency of these systems is also decreasing.
Integrating BlueField DPU into storage servers and host access can significantly improve storage efficiency by offloading and accelerating tasks such as NVMe oF storage protocol, thereby freeing up CPU cycles for other applications.
Support multi tenancyThe DDN Infinia storage platform adopts containerization, allowing different storage functions to run in separate containers, thereby achieving scalability and optimizing the entire data path, reducing latency by offloading tasks to DPUs.
Multi tenant deployment can integrate multiple namespaces within a single file system, thereby improving capacity utilization, reducing hardware costs, and simplifying deployment and management.
BlueField DPU is based on hardware isolation and resource allocation capabilities, allowing multiple users and applications to securely share infrastructure resources, thereby improving resource utilization and operational efficiency.
DDN Infinia provides local multi tenancy functionality for securely isolating user data; Apply efficient quality service algorithms to all tenants and sub tenants; And ensure data protection during transmission, static, and organizational sharing.
Enhance security
The dedicated processing resources and memory of BlueField DPU provide a secure environment that prevents unauthorized access and resists potential attacks. Hardware accelerated encryption ensures that data stored in the storage system is statically encrypted, thereby protecting sensitive information.
The access control mechanism of BlueField DPU enables administrators to define and execute fine-grained access policies to ensure that only authorized users or applications can access and modify data. In addition, a secure boot function is provided to verify the integrity of firmware and software components during the boot process, thereby preventing tampering or unauthorized modifications.
Uninstalling security related tasks from the host CPU can reduce the attack surface and free up CPU resources for other critical tasks.
With these combined security features, BlueField DPU provides a reliable and secure storage solution for AI workloads and data from DPU to CPU. The combination technology stack ensures that data is always protected, thereby addressing data security and integrity issues in AI driven environments. Your organization can provide better protection against cyber threats and unauthorized access, enhancing overall data security and compliance。
Enhance Expansion
DDN Infinia is a fully containerized platform built around a set of orchestrated microservices to provide the entire storage service. By using the BlueField data processor, DDN has developed a new architecture that supports a complete cloud native stack. This innovative use of BlueField data processors enables storage platforms to scale across networks.
Specifically, DDN Infinia’s Amazon S3 object service is containerized and can run independently of the Infinia storage system by utilizing the resources of NVIDIA data processors (DPUs) in the NVIDIA DGX client system. This design shift has completely changed the way data flows through storage systems. Traditionally, Amazon S3 object calls were made locally to services running on BlueField, but now this type of call can be made directly in NVIDIA DGX client systems, avoiding the latency of sending commands over the network (RESTful calls).
With BlueField, these calls are replaced by RDMA calls from DPU to the storage system. This will offload storage tasks from the main system and use more efficient data paths, significantly reducing latency and increasing bandwidth for artificial intelligence acceleration. The reconfiguration of this storage architecture has changed the data path of Amazon S3 objects, significantly improving performance and scalability.
SAGECONNECT TECHNOLOGY GROUP CO., LIMITED
Sageconnect is a reliable provider of IT network system integration and solutions for global enterprise customers.
