Dell Advances AI Data Platform to Boost Agent Use
TECHWORLD · · 1 views
✦ AI Summary
Dell Technologies is pushing ahead with upgrades to the Dell AI Data Platform.
On the 15th, it announced plans to add a unified semantic layer, an Enterprise Knowledge Graph and Knowledge Agents.
The planned release timing is the first half of 2027, and the data processing engine will be available in December.
Dell Technologies is pushing ahead with upgrades to the Dell AI Data Platform. The goal is to organize the meaning and relationships of dispersed internal enterprise data and increase the usability of AI agents. As enterprise AI deployment expands, data management has emerged as a key challenge alongside the models themselves.
Dell said enterprise data is spread across a range of environments, including files, databases and cloud services, and that a large share of existing data has not been classified or structured with AI agents in mind. In such a distributed and unstructured environment, Dell explained, AI agents must search multiple data sources for each query and re-confirm the meaning and relationships of the information.
Dell said this can increase the number of query steps, raise computing resource usage and slow response times. It added that costs may rise and the reliability of results could also be affected.
Accordingly, Dell Technologies announced on the 15th that it will add a unified semantic layer, an Enterprise Knowledge Graph and Knowledge Agents to the Dell AI Data Platform.
The Dell AI Data Platform manages the path from data storage locations to AI applications by linking storage, data orchestration, analytics and search functions. The platform combines storage with processing and retrieval capabilities to manage the flow from where data is stored to where it is used in AI applications.
The platform uses NVIDIA Nemotron Retriever to support document parsing, embedding and reranking. It also uses NVIDIA cuVS to accelerate vector indexing and search.
The newly added Unified Semantic Layer assigns consistent business meaning to structured and unstructured data. This connects terms such as "client" and "account," which are used differently across systems, as the same concept, and it can bring in and use existing ontologies and classification systems. Dell also supports the NVIDIA open-source library NVIDIA Auto-Ontology, which is used to build ontologies and knowledge structures based on enterprise data and expand the semantic layer.
The Enterprise Knowledge Graph links relationships between structured and unstructured data. Based on metadata, data lineage and query history, the structure continuously updates connections among data. As a result, when AI agents receive a question, they can simultaneously search tables, data products, multimodal data and vector indexes they are authorized to access. The point of this structure is to allow AI to reference relationships among data, not just search for individual pieces of information.
As a result, in situations such as a manufacturing site where anomalies occur, related information can be checked in connection with one another. If a sensor abnormality is detected, the repair history of the equipment can be checked together with it, and the production batches of suppliers can also be reviewed in relation to it. Potentially affected orders can also be checked in connection with it.
Knowledge Agents are based on domains defined in the knowledge graph. Knowledge Agents handle the function of providing information needed for specific tasks and topics. Administrators can set the data agents access, work guidelines, quality criteria and cost limits. Knowledge Agents also support reasoning using Nemotron Retriever and visual information processing.
The unified semantic layer, knowledge graph and Knowledge Agents are configured as a platform operating within enterprise data centers. Dell said the platform is designed to be independent of any specific AI model, data or storage vendor. It also said the platform is built to preserve existing enterprise data governance and access control systems.
Dell said it has also strengthened data processing acceleration functions and enhanced storage security and multitenancy capabilities. It added that it uses NVIDIA cuDF and Apache Arrow in the Dell Data Processing Engine. The engine accelerates data movement and analysis between storage and data processing systems.
As evidence of performance, Dell cited internal test results from last September. The internal testing compared the data processing engine using an NVIDIA RTX PRO 4500 Blackwell Server Edition GPU with CPU-based Apache Spark. Dell said the result was an average performance improvement of 3.9 times across several data processing workloads, and up to 20.4 times in batch processing workloads. However, actual performance may vary depending on system configuration and usage environment.
Dell Technologies said it has further strengthened Dell PowerScale's multitenancy and security features and refined role-based access control in the storage segment. These features are aimed at shared AI infrastructure environments used by multiple organizations and customers, and they support up to 500 tenants in a single cluster.
It also applied mutual transport layer security (mTLS) to NFS traffic to enable mTLS-based data encryption and authentication. This makes tenant-by-tenant data isolation management possible.
Dell Technologies also unveiled the Dell Storage Performance Tool. The tool measures S3-compatible object storage performance for training, inference and checkpointing tasks, and based on the results it can estimate the required infrastructure scale and compare storage environments.
In addition, Dell Technologies expanded its specialized services supporting AI data platform deployment. The service supports the setup of analytics, data processing, search and orchestration environments tailored to enterprise AI workloads, and also supports the transition to actual operations after deployment.
Dell presented the release timing for AI data platform features and storage product updates in stages. Dell said the planned release timing for the unified semantic layer, Enterprise Knowledge Graph and Knowledge Agents is the first half of 2027. The data processing engine with NVIDIA acceleration stack will be available in December.
Dell said the accelerated data processing function using Apache Arrow will be available in the first half of 2027. The planned release timing for the PowerScale security and multitenancy update is next month. The Dell Storage Performance Tool is available now.
Yusangmo, president of Korea Dell Technologies, said data without context is little more than noise. He also explained that data accessibility and actual usability are separate matters.
Yusangmo, president of Korea Dell Technologies, said the role of the Dell AI Data Platform is to give enterprise data consistent and reliable context and support the use of AI agents. He added that Dell will help enterprises maintain data governance and security, move AI quickly into operations and increase the value of existing data.
Source: TECHWORLD · Kim Seung-gi
Original: https://www.epnc.co.kr/news/articleView.html?idxno=408227
References
This article was produced with the help of an automated content generation algorithm.
Source: TECHWORLD
View originalThis article was summarized and organized by BizCrush based on the original article from TECHWORLD. For exact quotations and full details, please refer to the original article.