IEEE：2025企业生成式AI峰会报告-Day1（英文版）（46页）.pdf

1、August 20-21,2025San Jose,CAAugmenting GenAI Workloads on IBM Fusion HCIPurnanand Kumar/Dipali Chatterjee28 July 2025Augmenting GenAI Workloads with Content-Aware Storage on IBM Fusion HCI for Scalable,Trustworthy,and Accelerated Enterprise AIThe objective of the Content-Aware Storage(CAS)architectu

2、re is to facilitate smooth interactions between LLMs and extensive quantities of unstructured data,amplifying insight derivation and suggestion functionalities.IBM Fusion CAS uses IBM Storage Scale CNSA to remote mount a cluster where documents are parsed by NVIDIA NeMo services.Parsed content and m

3、etadata are embedded and indexed into a vector DB.Watch folders and optional AFM enable scalable,incremental ingestion.CAS provides semantic,keyword,and hybrid search APIs with optional reranking.These results can be integrated into enterprise RAG pipelines.Introducing IBM Content-Aware Storage(CAS)

4、:a software-defined storage data service that alleviates the knowledge base challenges for GenAI implementationsSimple:Automated RAG solutionEfficient:a)Cost/performanceb)Works with legacy dataEnables Gen AI capabilities on unstructured data in any on-prem locationOnly process incrementally changed

5、data;High performance shared storage for data processing;GPU optimized storage for optimized Document Processing and Search performance Preserve data ACL;Data encryption for embedding Secure:IBM CAS combines the power of AI document processing with IBMs AI storage software,&research innovationsto jo

6、intly bring to market a state-of-the-art storage-based Knowledge solution that isStorageGPFSNFSObject(S3)CAS Search APIFiles and DocumentsGen AI ApplicationsCHATRetreiveresponseQuery(prompt)Top k resultsGen AI AppsAgentQuery(prompt)Top k resultsresponseData&MetadataKnowledge BaseFile ProcessingData