The single most helpful thing about Dremio is how it completely changes the game on data movement. Before Dremio, our analytics team spent about 70 percent of their time just moving data around. We had ETL pipelines that took days to build, data copies sitting everywhere, and constant debates about which system should be the source of truth. Dremio just eliminated all of that. You connect it to your data sources, and you query them directly. No copying, no staging, no duplication. It sounds simple, but the impact has been enormous.
The user interface is clean and intuitive. It is web-based, which means no client installations to manage, and the layout is straightforward. The SQL editor is responsive with autocomplete and syntax highlighting, making query writing feel natural. The visual query builder is helpful for users who are not comfortable with SQL, though most of our team prefers writing queries directly. The dashboarding and visualization features are functional but not as polished as dedicated BI tools like Tableau or Power BI. For quick exploratory analysis, they work perfectly, but for polished executive presentations, we still export to other tools. Overall, the user experience is pleasant and focused on getting work done without unnecessary friction.
The integrations are where Dremio truly excels. The platform connects to over 36 different data sources, including all the major databases, data warehouses, and cloud storage platforms. We have it connected to PostgreSQL for transactional data, Snowflake for our warehouse, and S3 for our data lake files. The integration process is straightforward. You add a source, provide credentials, and the system catalogs the metadata automatically. No complex configuration required. The federated query engine means we can join data across these sources in a single query, which was impossible for us before. The platform also integrates well with BI tools like Tableau and Power BI through standard JDBC and ODBC connectors, allowing our analysts to work in the tools they already know.
The performance gains are genuinely impressive. When we first set up Dremio, we ran a test query that used to take about 8 minutes in our legacy environment. Dremio returned it in about 12 seconds. I actually thought it was broken. The Autonomous Reflections feature is magical. It learns what queries we run most frequently and automatically creates optimized data structures in the background. We do not have to think about performance tuning anymore. It just happens. Even with complex joins across multiple data sources, the query engine handles it efficiently. We have seen query times improve by 10x to 100x depending on the complexity of the workload.
The cost savings are another major win. We were paying a fortune for cloud data warehouse storage because we were duplicating everything. We had data in S3, data in Snowflake, data in Redshift, and we were paying for all of it. Dremio let us store everything in open formats on S3 and query it directly. We cut our storage costs by about 60 percent and our compute costs by about 40 percent. The open architecture is a big part of this. We are not locked into any proprietary format. If we decide to move away from Dremio, our data is still there in standard formats like Apache Iceberg. No vendor lock-in, no expensive migration project. The pricing model is consumption-based, so we only pay for what we use, which aligns well with our variable workload patterns. For organizations with significant data volumes, the ROI is substantial and measurable within the first year.
The onboarding experience was remarkably smooth compared to other enterprise platforms we have implemented. Dremio offers a Community Edition that we used to prototype and test before committing to the enterprise version. This allowed us to validate the platform's capabilities without any financial risk. The enterprise version includes formal support, and we have found the support team to be responsive and knowledgeable. Tickets are resolved quickly, and the documentation is thorough and well-organized. The platform is self-managed in our environment, which gives us full control, but Dremio also offers a fully managed cloud option if you prefer not to handle the infrastructure. The open-source community is active and helpful, which is a nice supplement to the formal support channels. Overall, getting started was straightforward and we were productive within weeks rather than months.
The AI Agent is a nice addition. It handles basic natural language questions well. A user can ask "Show me revenue by region for the last quarter" and it will generate the SQL and visualize the results instantly. It saves time on simple queries and helps non-technical users get started. The AI_GENERATE function is particularly interesting. It can extract structured information from unstructured files like PDFs or text documents directly within a SQL query, which opens up new use cases for us. The AI Semantic Layer ensures that both human analysts and AI tools are working with consistent business definitions, which improves the accuracy of AI-generated answers. However, the AI capabilities are not yet at the level where they can replace a skilled analyst for complex analytical questions. For day-to-day exploration and quick insights, it is genuinely helpful, but complex multi-condition analysis still requires human expertise.
Overall, Dremio has made us more agile, more cost-effective, and more data-driven as an organization. The reduction in complexity has been liberating. We are spending less time moving data and more time actually analyzing it. I would recommend it to any organization struggling with data silos, high costs, or slow access to insights. It is not perfect, but for what it does, it is remarkably effective. Review collected by and hosted on G2.com.
The most significant downside is that the platform often feels unfinished in production scenarios. It is not that the core functionality is broken, but the polish and robustness you expect from enterprise software isn't always there. Multiple reviewers have described that bugs surface frequently during actual use, which can be frustrating when you are trying to deliver reliable analytics to your business stakeholders.
Integrations are a particular sore spot. The connection with Power BI has been described as problematic by multiple reviewers, with customer support teams apparently having a poor understanding of how this integration works internally. This makes troubleshooting incredibly difficult when something goes wrong. The CI/CD capabilities also feel incomplete. There is an unsupported Python script available to handle development and production environments, but it cannot deal with some native Dremio objects like Row-Level Security policies. This means you might develop a dataset with proper access controls in your development environment, but the script simply will not deploy it to production, forcing you to find workarounds.
Performance at enterprise scale has been a documented weak point. The federation engine struggles with concurrency on large datasets, and standardized performance testing shows that the architecture routinely fails to complete queries when multiple users are running queries simultaneously. For AI use cases, this is more pronounced because an AI agent querying your data may be running dozens of complex federated queries in parallel as part of a single task, and the failure mode compounds quickly.
There are also technical limitations to be aware of. The platform enforces various system limits, such as maximums on autoingest pipes, sources, spaces, and total reflections. Query results returned via the console are limited to one million records, though you can use ODBC or JDBC to see the full result set. The platform also has limits on leaf columns in tables and queries, which can impact organizations working with wide datasets.
Support quality has been inconsistent. Some users have noted that support can be lacking at times, which they attribute to Dremio being a younger company compared to its competitors. The documentation around certain features, like Kubernetes deployment requirements, indicates significant administrative overhead, and managing the Kubernetes cluster requires considerable effort.
The AI capabilities, while promising, are somewhat lagging compared to other platforms. In a world where competitors have already integrated the latest AI techniques, Dremio is still catching up. The AI Semantic Layer, while useful for labeling data, does not enforce business rules. It organizes and labels data without packaging metadata, access rules, or business constraints into an enforceable framework. Different teams and AI agents can still interpret the same underlying data in incompatible ways, and the platform has no mechanism to stop them.
There is also a broader strategic concern. Dremio was recently acquired by SAP, which puts its roadmap in question. Features that Dremio was building independently now have to compete for budget and attention inside a much larger organization. Strategic shifts are already visible, with some cloud editions effectively put on hold while the deal works through approvals. For organizations with strict data governance requirements, the mandatory transmission of operational telemetry data back to Dremio's corporate endpoint introduces a compliance headache that cannot be ignored.
Despite these challenges, I still find Dremio valuable for our use case. The query performance is genuinely impressive, the cost savings are real, and the platform has transformed how we access data. The downsides are real and you need to go in with your eyes open. Budget extra time for troubleshooting integrations, be prepared for some administrative overhead, and have realistic expectations about the AI capabilities. If you have a focused use case like fast SQL acceleration on an Iceberg lakehouse where you own all the data, Dremio's core engine is still remarkably capable. But for teams needing broad federation across many sources, high concurrency, and polished enterprise features, it is worth evaluating carefully. Review collected by and hosted on G2.com.