English
Why combine Data Warehouse Automation and Data Test Automation?
Analytics Creator automates data warehouse design and deployment, while Big EVAL adds automated testing and data quality validation across the data pipeline. Together, they help teams build, deploy, and test data warehouse solutions faster, with less manual effort and better regression coverage.
Questions
- How does AnalyticsCreator automate data warehouse development?
- Why is testing still needed in automated data warehouse projects?
- How does Big EVAL test data pipelines and data quality?
- How can AnalyticsCreator metadata be used for automated testing?
- How does regression testing support agile data warehouse development?
Key Takeaways
- AnalyticsCreator is a design-time tool with no runtime lock-in.
- AnalyticsCreator generates source code and deployment artifacts for Microsoft environments.
- Big EVAL adds automated testing for data pipelines and production data validation.
- Testing remains necessary because source data, manual logic, and presentation layers can still introduce errors.
- Regression testing is important in agile sprint-based data warehouse development.
- Big EVAL can read AnalyticsCreator metadata and automatically apply test cases to repository objects.
- Data validation can be used across development, QA, and live environments.
- Both tools support automation across the development and quality assurance lifecycle.
Transcript
We have a short agenda with six sections. First, I will explain how AnalyticsCreator data warehouse automation works. Then Thomas will explain data warehouse testing and data quality testing. After that, Dimitri, our CTO, will present a live demo showing how to build a warehouse.
Because today’s session is limited to around 30 minutes, we will focus on the essentials. Longer versions of the demos are also available on YouTube. After the presentation, Thomas will demonstrate Big EVAL and explain how testing works. We will finish with a Q&A session and present a special offer for attendees.
It is important to understand how AnalyticsCreator works. AnalyticsCreator automates the design of a holistic data model and the complete processing flow from source systems through all warehouse layers to the frontend technology.
The design is independent of the target Microsoft technology. You can generate source code for SQL Server, Azure environments, tabular models, Power BI, Qlik, Tableau, Azure Data Factory, and SSIS.
The design and development processes are tightly connected. While designing the model, you immediately see the generated source code and can extend it with your own logic where needed.
We have many customer stories available on our website. One customer reduced time and costs in IT controlling by 80% using AnalyticsCreator. Another customer, mymuesli, successfully built its analytics environment after onboarding with AnalyticsCreator. A large SAP customer in the real estate sector reported achieving results 20 times faster than originally estimated.
Our vision is to give customers independence from software vendors and runtime dependencies. AnalyticsCreator is a pure design-time tool with no runtime component, meaning there is no vendor lock-in.
AnalyticsCreator generates the source code and deploys it into the Microsoft environment. After deployment, the generated code runs independently. Customers retain full ownership of the generated code and can use it freely.
We focus entirely on the Microsoft stack to deliver the best possible integration and results. The metadata repository is open, allowing customers and partners to build their own add-ons and extensions.
The process follows five simple steps. First, connect to the source systems. Next, extract metadata and define the data structures. Intelligent wizards then generate a draft warehouse model, which can be adapted with business logic and calculations. Finally, the solution is deployed to the target platform.
AnalyticsCreator supports many use cases, including new analytics platforms, modernization projects, Data Vault implementations, Synapse lakehouses, SAP integration, and exporting warehouses to other database platforms.
The main benefits are faster delivery, reduced risk, agile development support, and automation of repetitive development tasks. Developers can focus more on business requirements rather than repetitive coding.
AnalyticsCreator has also been recognized multiple times in the BARC Data Management Survey.
Nearly every second data record in business applications or external files contains at least one critical error. These issues can negatively affect reporting, analytics, and business decision-making.
According to Gartner research, poor data quality can create major financial impact. Even with automated warehouse generation, testing remains essential.
A fully automated factory still requires quality assurance. The same principle applies to data pipelines. Source data can contain errors, processes can fail, and transformations can introduce inconsistencies. This is why testing and quality management remain critical.
To achieve reliable analytics, two things are required: high-quality data and correctly functioning data processing systems.
If either the data or the processing logic contains errors, the result becomes “garbage in, garbage out.” This is why organizations need both data quality management and automated data testing.
Big EVAL covers both disciplines. For example, it can automatically verify whether all customer IDs from an ERP system exist in the data warehouse and continuously monitor this in production.
Even with automation, there are still many possible sources of error. Source systems may contain incorrect or incomplete data, users may enter unexpected values, and custom transformation logic may introduce mistakes.
One customer example involved point-of-sale systems occasionally failing to deliver data, resulting in incomplete warehouse loads.
Presentation layers and analytical models are also often built manually, which introduces additional risk. Even a simple SQL join mistake can cause major reporting problems.
This is why automated testing remains essential, even in highly automated environments.
Big EVAL supports testing across development, QA, and production environments. The same test cases can be reused throughout the entire lifecycle.
Quality gates can validate components before deployment, ensuring that incorrect data or broken logic does not reach production systems.
Regression testing is especially important in agile projects. Every sprint introduces new functionality, and all previous functionality must continue to work correctly. Automated testing ensures that increasing complexity does not reduce quality.
Big EVAL connects to many types of data sources through OLE DB, ODBC, REST APIs, flat files, and ERP systems such as SAP or Dynamics.
Test results are displayed in dashboards and can trigger alerts through email, Teams, Slack, Power Automate, Zapier, and other integrations.
Big EVAL also integrates directly with the AnalyticsCreator repository. It can automatically read metadata and generate test cases for warehouse objects based on the model itself.
Dimitri demonstrated how to create a new warehouse project in AnalyticsCreator using the Microsoft Northwind database.
AnalyticsCreator supports many connectors, including SQL Server, Excel, SAP, OData, Azure Blob Storage, and others.
A new connector was created for the Northwind database, and the connection was successfully established.
The Data Warehouse Wizard was used to select tables such as categories, suppliers, employees, customers, orders, and order details.
The wizard automatically imported the data, enabled historization, created dimensions, and generated a fact transformation.
AnalyticsCreator then generated the warehouse structure, including the source, staging, persistent staging, core, and data mart layers.
The staging layer imports the source data and supports filters, variables, and scripts for differential loading.
The persistent staging layer stores historized data using slowly changing dimension logic.
Additional validity columns such as “date from” and “date to” are automatically created, along with surrogate keys.
AnalyticsCreator supports multiple historization types, including SCD1 and SCD2. Developers can configure filters, variables, calculated columns, and scripts directly within the historization process.
Stored procedures for historization are automatically generated but can also be customized manually if required.
The core layer contains transformations, typically implemented as SQL views.
Developers can create regular transformations, manual SQL transformations, stored procedures, or external transformations using Integration Services or pipelines.
Fact transformations automatically combine related historized tables. AnalyticsCreator uses snapshot logic to simplify joins between historized datasets and supports historical point-in-time analysis.
Dimitri added a calendar dimension using reusable SQL macros. Macros simplify repetitive logic and dynamically generate SQL statements.
Measures such as quantity, unit price, and total price were added to the fact transformation.
The synchronize function then materialized the model into SQL Server and validated the syntax. Invalid logic was automatically detected during synchronization.
To improve performance, the fact transformation was persisted into a physical table.
AnalyticsCreator supports multiple materialization approaches, including full reloads, merge logic, and incremental loading.
The data mart layer acts as the reporting interface for Power BI and OLAP models. Measures and naming conventions can be managed directly within AnalyticsCreator.
AnalyticsCreator generates a Visual Studio deployment package containing everything needed to deploy the warehouse.
This includes DACPAC files, SSIS packages, Azure Data Factory pipelines, workflow packages, tabular models, and XMLA definitions.
Workflow packages orchestrate the execution order automatically. Once the deployment package is generated, AnalyticsCreator itself is no longer required for runtime execution.
The generated tabular OLAP cube included facts, dimensions, and measures ready for reporting.
The same model can be deployed both on-premises and in Azure environments, including Azure SQL, Power BI, and Azure Data Factory.
Thomas demonstrated the Big EVAL dashboard, where test results are grouped into test suites.
One example compared KPI values between the ERP system and the data warehouse. Another validated whether all customer records were successfully loaded into the warehouse.
Detailed test results and alerts can be sent directly to developers or data stewards.
Big EVAL test cases are created using configurable test algorithms and probes.
The platform supports comparisons, business rules, performance testing, and schema validation.
Tests can be executed manually, scheduled automatically, or integrated directly into ETL and deployment workflows.
Big EVAL can read metadata directly from the AnalyticsCreator repository.
One example automatically validated historized dimensions to ensure validity dates were correctly ordered.
Metadata-driven testing allows one test case to automatically scale across many dimensions and warehouse objects.
Big EVAL can compare source and target attributes across warehouse dimensions.
It dynamically reads metadata, generates queries, and validates whether data has been transformed correctly.
Business rule validation is also supported. Examples include validating employee data completeness or checking age-based vacation rules.
Records violating rules can automatically be sent to data stewards for correction.
Big EVAL integrates with webhooks, Slack, Teams, Power Automate, and Zapier to trigger alerts and workflows automatically.
Special attendee offers included discounted Big EVAL subscriptions and AnalyticsCreator packages, along with a free online trial.
AnalyticsCreator supports collaborative development through shared repositories.
Projects can be stored as SQL files and managed with Git, Bitbucket, or Subversion. Developers can lock specific warehouse objects or object groups to avoid conflicts.
Partial deployment packages can also be created for specific project areas.
Repositories can additionally be stored in the AnalyticsCreator cloud, where all previous versions remain accessible.
Big EVAL is technology-agnostic and supports relational databases, NoSQL systems, flat files, data lakes, OLAP cubes, and many other technologies.
It can also test security concepts by impersonating Active Directory users and validating access permissions.
Compared to custom-built testing frameworks, Big EVAL provides a standardized and scalable approach that can be used immediately without internal development effort.
AnalyticsCreator trials can be requested directly from the website. Users receive onboarding instructions and support during the evaluation phase.
For Salesforce connectivity, AnalyticsCreator supports both third-party connectors from CData and OData-based integration approaches.
