From file systems to AI insights: Dremio Cloud + Amazon FSx for NetApp ONTAP

Every enterprise has a data problem hiding in plain sight. Not the kind that shows up in board decks about cloud migration or AI strategy. The quieter kind: petabytes of files sitting in NetApp ONTAP systems — financial records, engineering documents, customer data, sensor logs — that power daily operations but stay invisible to every analytics tool and AI model the company has invested in. 

The conventional fix is a migration project. Move the data to S3, build ingestion pipelines, wait six months, hope nothing breaks. It works, eventually. But when leadership is asking for AI ROI this quarter, not next year, that timeline doesn’t hold up. 

There’s a faster path. NetApp and Dremio have built a joint solution on AWS that turns existing ONTAP file data into a governed, AI-ready analytics layer — without moving it out of your file systems. 

Enterprises have spent decades building workflows around NetApp ONTAP. Manufacturing lines write quality data to NFS shares. Healthcare systems store patient records on SMB volumes. Financial services firms keep trading data in ONTAP clusters that have been running since the Obama administration. 

That data isn’t going anywhere — and it shouldn’t. These systems work. The applications that depend on them work. The governance and access controls around them work. 

The problem is that AWS’s most powerful AI and analytics services — Bedrock, SageMaker, Athena, Glue, Redshift — all speak S3. They can’t reach into an NFS mount or an SMB share. So organizations end up building one-off pipelines, copying subsets of data into S3 buckets, and managing a growing tangle of duplication that nobody trusts and everybody maintains. 

80% of enterprise data is unstructured, stored across file systems worldwide. Most of it has never been accessible to a single analytics query. 

In December 2025, AWS and NetApp launched S3 Access Points for Amazon FSx for NetApp ONTAP. The concept is straightforward: attach an S3 endpoint to your ONTAP volume, and every file in that volume becomes accessible through the S3 API. No copies. No pipelines. No migration. Just secure, always up-to-date access, for modern analytics tools and AI applications.  

Setup takes minutes. You give the access point a name, pick a file system identity (UNIX or Windows), and define who can access it. The access point gets a unique alias that any S3-compatible service treats as a bucket name. Your files stay exactly where they are, still accessible through NFS and SMB, but now also available to the 50+ AWS services built around S3. 

For organizations already running ONTAP on-premises, SnapMirror handles the bridge to AWS. Block-level replication with built-in deduplication and compression moves data to FSx for ONTAP in AWS. Once it’s there, the S3 Access Point lights up instantly. 

That’s the storage story. But access alone doesn’t give you analytics. 

Dremio Cloud connects to FSx for ONTAP through its S3 Access Point and treats it like any other data source in the lakehouse. But Dremio does more than just read the data. It builds a governed, performant analytics layer on top of it — one that serves both human analysts and AI agents. 

Here’s what that looks like in practice: 

Read in place, write to ONTAP. Dremio reads your existing file data through the S3 Access Point and then writes processed, curated data back into native FSx for ONTAP S3 buckets. You can run a medallion architecture — bronze, silver, gold — entirely within ONTAP, backed by its enterprise-grade data management: snapshots, clones, tiering, and replication. 

Federation that actually works at scale. Most federation tools demo well but fall apart under production load. Dremio solves this with Reflections — smart materializations that the platform creates and maintains automatically. Dremio learns from query patterns and pre-computes the results that users and agents will need, rewriting incoming SQL at runtime to hit the materialization instead of scanning raw data. The result: sub-second performance on federated queries without manual tuning. 

AI-native from the ground up. Dremio’s AI Agent lets anyone ask questions in plain English and get answers backed by governed data. The platform generates SQL, validates results, and delivers insights with visualizations — no coding required. For teams building their own agents, Dremio’s MCP Server (Model Context Protocol) gives any AI model — Claude, ChatGPT, or custom — direct, governed access to the full data environment. 

Unstructured data processing in SQL. Dremio’s AI Functions bring large language models directly into SQL queries. AI_GENERATE converts PDFs, documents, and images stored in ONTAP into queryable structured data. AI_CLASSIFY runs sentiment analysis and categorization. AI_COMPLETE handles summarization. All of this runs inside the lakehouse, against data in FSx for ONTAP, with no separate ML pipeline to build or maintain. 

Fair question. If S3 is required for the most powerful analytics and AI tools, why not copy it all into plain S3 buckets? 

For some workloads, that makes sense. But for organizations with significant ONTAP investments, the answer usually comes down to three things. 

Your data is already there. If petabytes of unstructured data sit on ONTAP today, an S3 Access Point gives you instant access without a migration project. No project plan, no data validation phase, no cutover risk. It’s available now. 

ONTAP does things S3 doesn’t. Snapshots, clones, cross-region replication with SnapMirror, storage tiering, ransomware protection, WORM compliance — these are capabilities that enterprises rely on, and that native S3 can’t provide. Running your analytics layer on ONTAP means you don't have to choose between enterprise capabilities and accelerated insights. 

Dual-protocol access. Your existing applications keep reading and writing through NFS and SMB. New analytics and AI workloads access the same data through S3. One copy of the data, secure future-proof access, and zero disruption to existing workflows. 

The full setup — from ONTAP filesystem to running AI-powered analytics — takes an afternoon, not a quarter.

1. Deploy your FSx for ONTAP filesystem in AWS if you haven’t already. If you’re running ONTAP on-premises, this is your cloud landing zone.
2. SnapMirror your data from on-premises ONTAP to FSx for ONTAP. Block-level replication with dedupe and compression keeps the transfer efficient.
3. Create an S3 Access Point on your FSx for ONTAP volume. This takes minutes and requires no application changes.
4. Deploy Dremio Cloud with a free $400 trial. No infrastructure to provision — it’s fully managed.
5. Connect Dremio to your FSx for ONTAP via the S3 Access Point. Start querying, building semantic models, and running AI agents against your data.

The gap between “we have the data” and “we have AI-driven insights” has been measured in months and millions of dollars. For organizations running NetApp ONTAP, that gap just collapsed in an afternoon.

Your data doesn’t need to move. It needs a smarter way to be found, understood, and put to work.

• Deploy FSx for ONTAP 
• SnapMirror to AWS 
• Create S3 Access Point 
• Start Dremio Cloud 
• Connect Dremio to S3