How Dobler Insights works
Dobler Insights takes the file or database you have today and turns it into the dashboard your people open. It works in four layers, ingestion, storage, security and delivery, with source control around all of it.
Files go in as spreadsheets and come out as structured tables.
Drop a file into a monitored location, such as Azure Blob Storage, a SharePoint document library or an SFTP folder, and the engine picks it up.
- Formats: CSV with any common delimiter, and Excel XLS and XLSX, including workbooks with several sheets. Character encoding is detected automatically.
- Structure checks, three ways. Strict mode rejects any file that does not match the expected columns and types. Adaptive mode adds new columns and fills missing ones with blanks. Discovery mode reads the structure from the first file, then switches to strict or adaptive.
- All or nothing. A file loads completely, or the previous version stays in place.
- A record of every load: when it ran, the file's fingerprint, the row count and the result of every check.
Databases are copied into daily snapshots, without touching production.
Dobler Insights connects to your databases, including SQL Server, Oracle, MySQL, PostgreSQL, Azure SQL, Azure Synapse, Amazon RDS, Google Cloud SQL, Snowflake, Databricks and BigQuery, as well as REST, OData and SOAP services.
- Read-only. Extraction uses read-only connections, and credentials are fetched from Azure Key Vault at run time, never stored in configuration or code.
- A daily snapshot. Each run stores a complete copy of the source as it stood at that moment, and that copy never changes. Reports read the snapshot, never the live system, so reporting does not slow down or lock your operations.
- Checked before it is used. Row counts, missing values and relationships are checked on every run. A large deviation raises an alert instead of failing silently.
History keeps what the numbers looked like on any date.
The daily snapshots build up into a history of your data. Where only the latest value matters, such as a customer's email address, only the latest is kept. Where the change matters, such as a customer's tier, an employee's department or a product's price, every version is kept with the dates it was true.
That makes three kinds of question easy to answer: how this period compares with the last one on a consistent basis, what a figure looked like on a particular date, and exactly what data existed when a decision was made.
Storage on Microsoft Fabric.
Ingested data is held in Lakehouse and Warehouse storage on Microsoft Fabric, the foundation the rest of Dobler Insights is built on.
Security works at three levels.
- Sign-in. People sign in to the Dobler Insights portal with Microsoft Entra ID, SAML, or another identity provider.
- Report. Each person's permissions say which workspaces and reports they can open, and they are checked before any report loads.
- Data. Row-level security in the data model limits each person to their own rows. It cannot be bypassed by editing a report, using bookmarks or changing a URL.
Delivery is Power BI, without a license per viewer.
Dashboards, reports and paginated reports reach your people through your branded Dobler Insights portal, and none of them needs a Power BI license. Paginated reports cover the documents that must print exactly right, such as financial statements, invoices and compliance reports, with export to PDF, Excel and Word.
Every change is controlled.
Every report, dataset and deployment setting is kept in Git. Changes move from development to user testing to production, with a check at each step. Emergency fixes are possible, and they are logged and approved like everything else.
The platform is operated with Fabric Control.
Fabric Control, the Dobler Data Solutions control plane for Microsoft Fabric, handles monitoring, backup and capacity management for the platform, and logs every administrative action, configuration change and deployment.
See it on your own data.
Bring one source, and we will show you the whole path from the file to the secured dashboard, including the parts that take real work.