Data Engines
Open, modular engines that query, search, enrich, and process your structured and unstructured data for AI.
From raw data to consumable data products
Data Engines unify data preparation and access on Dell S3 object storage to ingest and transform data, and delivering consumable, AI-ready data products.
Purpose-built engines for search, analytics, and processing
Unlike proprietary engines that limit flexibility, Data Engines are built with open, cutting-edge tools to handle the complexity of modern data, integrated directly for data engineers and architects.
Open, federated, and built on trusted storage
Three choices separate Data Engines from proprietary, lock-in analytics stacks.
Open, best-of-breed
Data Engines are built on open, cutting-edge tools—such as Starburst, Elastic, and Apache Spark—and open table formats. Avoid the lock-in of proprietary engines and integrate the tools you already trust.
Query where data lives
Federated access lets you search and analyze structured and unstructured data across cloud, on-premises, databases, lakes, object stores, and NoSQL without moving or duplicating data into a separate platform first.
Trusted enterprise storage
Data Engines run on Dell S3-compatible storage such as ObjectScale and PowerScale, delivering the performance, reliability, and security AI demands at up to 76% lower total cost than public cloud object storage.
Questions
What are Data Engines?
Data Engines are the data-processing layer of the Dell AI Data Platform with NVIDIA, a set of open, modular engines for search, analytics, and large-scale data processing. The engines query, search, enrich, and transform structured, semi-structured, and unstructured enterprise data so the data is ready for analytics, business intelligence, and AI workloads like retrieval-augmented generation (RAG) and agentic retrieval. Only adopt the engines you need, run on your existing Dell storage, without moving data into a proprietary platform.
What is the Data Analytics Engine, and how does Starburst fit in?
The Data Analytics Engine, powered by Starburst, is a federated query and analytics engine that lets you analyze data across many sources—such as data lakes, warehouses, object stores, and databases—without first copying data into one location. The engine uses open table formats such as Apache Iceberg and runs SQL analytics directly where your data lives, so teams get faster insight with less data duplication and lower analytics cost.
Can I run analytics where my data already lives?
Yes. The Data Analytics Engine federates queries across distributed sources, so you can analyze data in place instead of building fragile extract, transform, load (ETL) pipelines to move data into a central warehouse. This reduces copies, cuts storage and pipeline costs, and shortens time to insight while keeping data under existing governance and access controls.
How do Data Engines support vector search and RAG?
The Data Search Engine, built on Elasticsearch and accelerated by NVIDIA cuVS, generates embeddings and builds vector indexes so enterprise data can be retrieved semantically for RAG and agentic AI. GPU acceleration delivers up to 12 times faster vector indexing throughput, keeping indexes fresh as data changes so retrieval stays fast, relevant, and grounded in current enterprise context.
Can Data Engines search both structured and unstructured data?
Yes. The engines are designed for structured, semi-structured, and unstructured data, including documents, images, video, logs, records, and other enterprise content. Search combines keyword, semantic, and hybrid retrieval across these data types, so users and AI applications can find the right information regardless of where data lives or what that data takes.
How do Data Engines prepare and transform data at scale?
The Data Processing Engine, built on Apache Spark, handles large-scale ingestion, cleansing, transformation, and enrichment. The engine prepares raw enterprise data (batch or streaming) into clean, structured, AI-ready datasets. Teams can apply GPU acceleration to speed data preparation ahead of training, indexing, and inference so pipelines keep pace with changing data.
What do I actually deploy to run Data Engines?
Data Engines run on a validated foundation such as the Data Analytics Engine appliance (DDAE660) together with Dell storage—PowerScale for scale-out file and ObjectScale for S3 object—and NVIDIA acceleration. Start with a single engine for a specific problem and expand across search, analytics, and processing as your workloads grow, without re-architecting.
How do Data Engines help lower analytics and storage cost?
By querying where data lives, the engines reduce redundant copies and ETL overhead, and open table formats avoid proprietary lock-in. Combined with cost-efficient Dell storage, customers can significantly lower analytics and object-storage costs while improving query performance, getting more insight per dollar instead of scaling cost linearly with data.
How do Data Engines fit with the rest of the AI Data Platform?
Data Engines are the search, analytics, and processing layer of the Dell AI Data Platform with NVIDIA. They work alongside Data Orchestration (which automates the end-to-end pipeline) and Dell Storage for AI (the AI-ready storage foundation). Together they take enterprise data from ingest to AI outcome. Because the platform is modular, you can start with Data Engines alone and add the other layers when needed.