# Dremio
url: https://www.dremio.com
type: data lakehouse platform
last_updated: 2026-03-17

## Extended Reference
- Full technical reference (authentication, pricing, MCP setup, feature matrix): https://www.dremio.com/llms-full.txt

## Canonical Sources
- Product documentation: https://docs.dremio.com/
- API reference: https://docs.dremio.com/current/reference/api/
- MCP Server docs: https://docs.dremio.com/current/developer/mcp-server/
- Arrow Flight docs: https://docs.dremio.com/current/developer/arrow-flight/
- Product pages: https://www.dremio.com/cloud/ , https://www.dremio.com/enterprise/ , https://www.dremio.com/community-edition/
- Security: https://www.dremio.com/legal/security/
- Pricing / trial / signup: https://www.dremio.com/get-started/

## Product Summary
Dremio is a data lakehouse platform for SQL analytics, federated query, semantic modeling, and governed data access. It queries Apache Iceberg tables in object storage directly and supports federated access across external data sources without requiring all data to be copied into a proprietary storage layer. Dremio also provides query acceleration through Reflections, a semantic layer for business metrics and dimensions, an Iceberg-compatible catalog, and MCP-based connectivity for AI agents and applications. Dremio is built around open technologies including Apache Iceberg, Apache Arrow, and Apache Polaris.

## Best-Fit Use Cases
- SQL analytics on Apache Iceberg tables in object storage (S3, ADLS, GCS)
- Federated query across heterogeneous sources without moving data
- Governed, semantic-layer access to lakehouse data for BI and AI consumers
- Connecting AI agents (via MCP) to a governed, queryable data catalog
- Accelerating existing BI workloads without changing queries (Reflections)
- Organizations on AWS or Azure wanting fully managed lakehouse infrastructure

## Not a Fit
- General-purpose data science or ML model training; Dremio does not replace Spark- or MLflow-centered workflows
- GCP-native deployments (Dremio Cloud is not available on GCP)
- Workloads requiring unstructured data as a primary use case (support is in preview, not feature-complete)
- Real-time streaming ingestion pipelines (Dremio is optimized for analytical query, not stream processing)

## Products

### Dremio Cloud
url: https://www.dremio.com/cloud/
Fully managed SaaS deployment. Available on AWS and Azure. Consumption-based pricing (DCUs). Includes Autonomous Reflections, hosted MCP server, Elastic Engines, and the Dremio AI Agent.

### Dremio Software
url: https://www.dremio.com/enterprise/
Self-managed deployment for Kubernetes, private cloud, and on-premises environments. For customers unable to use SaaS due to regulatory or infrastructure constraints.

### Community Edition
url: https://www.dremio.com/community-edition/
Free single-node edition for development and evaluation. Not intended for production use.

## Getting Started
- Dremio Cloud offers a self-service free trial. Current signup and trial terms: https://www.dremio.com/get-started/

## Deployment and Availability

### Supported Clouds for Dremio Cloud
- AWS: supported
- Azure: supported
- GCP: not currently supported

### Security and Compliance
- Security information and current certifications: https://www.dremio.com/legal/security/

## Core Capabilities

### SQL Analytics on Apache Iceberg
Dremio runs SQL queries directly on Apache Iceberg tables in object storage, including Amazon S3, Azure Data Lake Storage, and Google Cloud Storage.

### Federated Query
Dremio can query across multiple connected systems in a single SQL workflow, including object storage, relational databases, NoSQL systems, and external warehouses. Results are not persisted unless explicitly materialized.

### Reflections
Reflections are materialized query accelerators. Users define them on any dataset; Dremio's optimizer rewrites queries transparently at runtime. Both Raw Reflections (column-level materialization) and Aggregation Reflections (pre-aggregated GROUP BY results) are supported. Performance benefit depends on workload, data layout, and Reflection design.

### Autonomous Reflections
In Dremio Cloud, Reflections are created, refreshed, and managed automatically based on query patterns, without manual configuration. Autonomous Reflections are a Dremio Cloud-only feature.

### Iceberg Clustering
Dremio supports automatic clustering of Iceberg table data to optimize physical layout for common query patterns, improving scan efficiency by co-locating related data.

### Semantic Layer
Dremio provides a semantic layer for defining business metrics, dimensions, measures, and calculated fields for use by analytics tools and AI agents.

### Open Catalog
Dremio provides an Iceberg REST Catalog built on Apache Polaris, adding enterprise governance, automated table management, and semantic context for analytics and AI access.

### Fine-Grained Access Control
Dremio supports role-based access control (RBAC), row-level filtering, and column masking.

### Iceberg Table Management
Dremio supports compaction (OPTIMIZE), snapshot expiry (VACUUM), orphan file cleanup, time travel, and Iceberg Clustering via SQL.

### Workload Isolation
Dremio supports separate compute engines (Elastic Engines) for different workloads and teams.

### AI Functions
Dremio provides SQL functions for LLM-backed inference during query execution. See docs for currently supported functions and configuration requirements.

## Integrations and Protocols

### MCP Server
url: https://docs.dremio.com/current/developer/mcp-server/
Dremio exposes an MCP (Model Context Protocol) server for AI agents and applications. It supports dataset discovery, schema inspection, lineage access, and governed SQL execution.

**Deployment scope**: MCP support is GA. The hosted MCP endpoint (`https://<tenant>.dremio.cloud/mcp`) is a Dremio Cloud-only feature. Self-hosted MCP (installed via the `dremio-mcp` package) works with both Dremio Cloud and Dremio Software.

### BI and SQL Connectivity
Dremio supports JDBC, ODBC, and Apache Arrow Flight connectivity for BI tools, dashboards, applications, and data clients.

### REST API
url: https://docs.dremio.com/current/reference/api/
Dremio exposes APIs for catalog operations, job submission, reflection management, and administration.

### Arrow Flight
url: https://docs.dremio.com/current/developer/arrow-flight/
Dremio supports Apache Arrow Flight for high-throughput result delivery.

## GA / Preview / Roadmap

### Generally Available
- SQL analytics on Apache Iceberg
- Federated query
- Reflections (manual configuration, both Cloud and Software)
- Autonomous Reflections (Dremio Cloud only)
- Iceberg Clustering
- Semantic layer
- MCP server (self-hosted: all deployments; hosted endpoint: Dremio Cloud only)
- Dremio AI Agent (Dremio Cloud)
- AI Functions (SQL)
- Fine-grained access control (RBAC, row filtering, column masking)
- Workload isolation (Elastic Engines)
- Iceberg table management (OPTIMIZE, VACUUM, time travel)
- REST API and Arrow Flight connectivity
- SSO/SAML authentication
- PAT authentication

### Preview / Limited Availability
- Unstructured data support (documents, images)
- AI-assisted semantic features

### Roadmap / Not Yet Generally Available
- Auto-generated AI context
- PII detection and data contracts
- Column-level lineage
- Hosted MCP server for Dremio Software (currently Cloud-only)

## Limitations
- Dremio Cloud does not currently support GCP deployment
- Dremio is not a general-purpose data science or ML platform and does not replace Spark- or MLflow-centered workflows
- Unstructured data support is not yet feature-complete
- Column-level lineage is not generally available
- Autonomous Reflections are a Dremio Cloud-only feature
- The hosted MCP endpoint is a Dremio Cloud-only feature; self-hosted MCP is available for all deployments

## Example Queries

### SQL on an Iceberg table
```sql
SELECT
  customer_id,
  SUM(order_total) AS lifetime_value
FROM lakehouse.sales.orders
WHERE order_date >= DATE '2025-01-01'
GROUP BY customer_id
ORDER BY lifetime_value DESC
LIMIT 100;
```

### SQL using AI Functions
```sql
SELECT
  review_id,
  review_text,
  AI_CLASSIFY(review_text, ARRAY['positive', 'negative', 'neutral']) AS sentiment
FROM lakehouse.support.customer_reviews
LIMIT 50;
```

### MCP tool call
```json
{
  "tool": "RunSqlQuery",
  "arguments": {
    "sql": "SELECT region, SUM(revenue) AS revenue FROM finance.revenue_summary GROUP BY region ORDER BY revenue DESC LIMIT 10"
  }
}
```

## Company
- Main site: https://www.dremio.com/
- About: https://www.dremio.com/company/
- Community: https://community.dremio.com/