One platform for your data and Power BI, with workspaces, permissions and capacity set up correctly from the start.
Trusted by leading organisations
A Microsoft Fabric implementation brings OneLake, lakehouses or warehouses, pipelines and Power BI models together in a managed environment, with clear arrangements for capacity, permissions and deployment. Data Science Lab sets up this environment for organisations already using Microsoft 365, Azure or Power BI that want to replace fragmented reporting with a single reliable foundation.
It often starts with a trial capacity and an enthusiastic analyst. Before long, there are workspaces called 'Test Jan' and 'Sales nieuw', three Dataflows Gen2 each bring in the same ERP tables in their own way, and every Power BI report has its own semantic model. At the revenue meeting, two different figures are on the table.
During the month-end close, a resource-intensive refresh runs on the same capacity as the controller’s reports. Fabric spreads background consumption over a 24-hour period, but an overloaded capacity first slows down new interactive requests and then rejects them. The controller sees a report that will not load.
The technology is rarely the cause. What is missing is a structure for workspaces, an owner for each dataset and a defined route to production. That is as much a question of data governance as it is of engineering.
OneLake provides one logical data lake per tenant. We organise it into domains and workspaces by department or data layer, making ownership clear and allowing teams to use each other’s tables through shortcuts without copying them.
If your team works with Spark and notebooks, a lakehouse is a natural choice; for T-SQL with transactions across multiple tables, a warehouse is a better fit. A common setup, also described in Microsoft’s documentation, uses a lakehouse for the bronze and silver layers and a warehouse for the gold layer.
Data Factory pipelines orchestrate the loading steps, Dataflows Gen2 serve those working with low-code, and notebooks handle the heavier transformations. Mirroring keeps sources such as Azure SQL Database, SQL Server and Snowflake almost in sync with OneLake in real time.
A semantic model in Direct Lake mode reads Delta tables from OneLake without a full import refresh. This suits the gold layer, provided the tables are well maintained and stay within your capacity limits. Read more about Power BI dashboards.
We distribute workspaces across F-SKU capacities, monitor usage in the Capacity Metrics app and enable surge protection so background tasks do not block other people’s reports. You can pause a capacity when it is not in use.
Workspace roles assigned through groups, OneLake security down to row and column level, and sensitivity labels from Microsoft Purview determine who can see what. Git integration with Azure DevOps or GitHub and deployment pipelines move changes into production in a controlled way.
We’ll look at your data sources, reports and licences together and discuss an initial use case.
We map data sources, Power BI workspaces, semantic models, gateways and capacity usage, so you can see what can be reused and what needs to be rebuilt.
We choose between a lakehouse and a warehouse for each layer, define domains and workspaces, and agree who owns each dataset.
We build one application from source to report, moving through bronze, silver and gold layers to a semantic model. Users compare the results with their current figures.
From that first use case, we work with Git integration, deployment pipelines and separate workspaces for development, testing and production, with access permissions managed through groups.
We monitor interactive and background usage over a full monthly cycle, including month-end closing, then adjust capacity or how workspaces are distributed.
Your team gets the documentation and guidance to continue building on the platform themselves, or we manage and develop it after launch.
Working with DSL has enabled us to turn our data into actionable steps, leading to targeted and effective campaigns and activations.
Fabric is a SaaS platform within your Microsoft tenant. It works best if Power BI is already your standard, your data sources are mainly in Microsoft services or common databases, and your team prefers working with SQL, low-code and notebooks over managing clusters. We cover it as part of our data foundation service.
Fabric builds on what you already have. Power BI Premium workspaces can move to an F-SKU, you can mirror databases instead of reloading them, and we migrate work from Azure Synapse one object at a time. The old and new environments run side by side until their results match.
If your focus is on data engineering and machine learning in Python and Spark, you work on AWS or Google Cloud, or you already have a mature Databricks environment, also read about our approach to Databricks implementation. The two are not mutually exclusive: Fabric can mirror the Azure Databricks catalogue, allowing Power BI to read those tables without a copy.
Any organisation can start an AI pilot. The challenge is what comes next: making AI part of the organisation. That’s where Data Science Lab makes a difference.
Read more about us
10+
8.7
200+
We set up the components you actually use: OneLake with domains and workspaces, lakehouses or warehouses for each data layer, pipelines or mirroring to load data, and semantic models for Power BI. We also configure permissions, sensitivity labels, Git integration and deployment pipelines. You get an initial working use case on a platform your team can expand without starting again.
Microsoft no longer sells new P-SKUs, and existing subscriptions end when their contracts expire. You then move your workspaces to a Fabric F-SKU, preferably in the same region, while reports and apps continue to work. Check the licensing: on F64 and above, viewers with a free licence can view content; on smaller F-SKUs, every viewer needs a Pro or PPU licence.
Yes, but expect to copy and adapt things rather than move everything with one click. Microsoft provides migration assistants for dedicated SQL pools and Spark notebooks. Some data types change: datetimeoffset does not exist in the Fabric warehouse, so time zone information goes into a separate column. We migrate in small steps and compare the results with the old environment.
By measuring before scaling up. The Capacity Metrics app shows which tasks are consuming capacity, and the cause is often in the design: a dataflow that loads too much data or Delta tables with too many small files. Surge protection rejects background tasks before interactive users are affected, and critical reporting can run on its own capacity.
We grant access through groups in the Admin, Member, Contributor and Viewer workspace roles. Within OneLake, security roles restrict access to folders, tables, rows or columns, and those rules apply in SQL, Spark and Power BI. Sensitivity labels and DLP policies from Microsoft Purview protect sensitive data, while the audit log shows who did what.
We are a Microsoft Partner and have helped organisations with data and AI since 2016. For See Tickets, we brought ticketing and behavioural data together on a scalable Azure platform for recommendation models that learn continuously. For Ammega, we built Power BI dashboards and a Power BI app for management, planning and finance. We also place data platform engineers and BI specialists with clients.
If your reporting runs on a few Power BI models that work well and no new teams or sources are being added, Fabric mainly adds management overhead. In that case, cleaning up the data model or improving a source connection will help more. If you already have a well-functioning lakehouse on Databricks, connecting Power BI to it is often enough. We’ll tell you honestly when a smaller step will do.
E-mail info@datasciencelab.nl
Phone +31 (0)20 3636 724
Tell us what sources, reports and licences you currently have. We’ll help you think through a first use case and what it takes to put it into production.
Message sent
Thanks for your message. We’ll be in touch as soon as possible.
