To accelerate customer AI pipelines and workloads, Dell has integrated its AI Data Platform with Elasticsearch. At the same time, the PowerEdge R7725 server runs the Nvidia RTX Pro 6000 Blackwell GPU and incorporates Nvidia’s AI Data Platform Reference Design.
The Dell AI Data Platform will soon have an unstructured data engine, built by Dell and Elastic, with vector search, semantic retrieval and hybrid keyword search capabilities. It will provide real-time, secure access to large-scale unstructured datasets for inferencing, analytics, and intelligent search, using GPU acceleration. The unstructured data engine works alongside other tools, like a federated SQL engine for querying scattered structured data, a processing engine for handling large-scale data transformation, and storage designed for fast, AI-ready access.
Ken Exner, chief product officer at Elastic, stated: “With Elasticsearch vector database at the heart of the Dell AI Data Platform’s unstructured data engine, Elastic will bring vector search and hybrid retrieval to a turnkey architecture, enabling natural language search, real-time inferencing, and intelligent asset discovery across massive datasets. Dell’s deep presence in the enterprise makes them a natural partner as we work to help customers deploy AI that’s performant, precise, and production-ready.”
GPU-powered search can use the R7725 server, which Dell says is “ideal for accelerated enterprise workloads, from visual computing, data analytics and virtual workstations.” Typically for this line of work, the servers will be fitted with SSDs – 1.6 TB to 30.72 TB NVMe as well as a range of SATA ones. SAS disk drives are also available. The servers can be linked to Dell’s external PowerScale (scale-out NAS) and ObjectScale (S3-compatible object storage) and their high-capacity stores of structured, semi-structured, and unstructured data.
As sister pub El Reg said, the RTX Pro 6000 is a workstation GPU, puts out 3,753 teraFLOPS of sparse FP4 compute, and is equipped with 96 GB of GDDR7 with 1.6 TBps of memory bandwidth. Nvidia’s AI Data Platform reference design integrates storage with its hardware and software to enable AI agents to deliver real-time responses. The Nvidia components are Blackwell GPUs, BlueField-3 DPUs, Spectrum-X networking, and Nvidia AI Enterprise software. The software bit includes NeMo Retriever and NIM microservices, and the AI-Q Blueprint.
The R7725 and also Dell’s R770 servers can run Nvidia AI reasoning models such as the latest Nemotron models for agentic AI and Cosmos world foundation models for physical AI.
All this enables Dell to claim that this update to its AI Data Platform will “enhance unstructured data ingestion, transformation, retrieval and compute performance to streamline AI development and deployment – turning massive datasets into reliable, high quality real-time intelligence for generative AI.”
This pair of updates “will help support enterprises on their AI journey from data ingestion to inferencing, streamlining data preparation, unifying data access across silos and delivering end-to-end enterprise-grade performance.”
The unstructured data engine in Dell AI Data Platform will be available later this year. Dell PowerEdge R7725 and R770 servers with Nvidia RTX Pro 6000 GPUs will be globally available later this year.
Read more in a Dell blog.
