Open standards are an important way to become more digitally independent. They make it easier to switch applications and give organisations more control over their own systems.
That is why DataOwn uses open standards wherever possible. For data transformations, we use SQL: a language that many BI platforms understand. This is an advantage over solutions where the logic is mainly built in a specific drag-and-drop tool.
Yet Business Intelligence still lacks interoperability in many areas. There is no standard for dashboards, including their charts, layouts and interactions, so they cannot readily be exchanged between tools. There are also still no widely adopted standards for the semantic layer — the place where you define metrics, dimensions and relationships between tables. As a result, the semantic layer and the dashboard are often closely connected. Organisations end up choosing one platform for both, even when they would prefer to choose those components separately.
Apache Ossie could change that. Ossie is an open standard for exchanging semantic models. The aim is to reuse definitions across different analytics and BI tools. For example, you could manage your semantic layer in one tool and your dashboards in another. That ideal has not yet been achieved: the standard and support from different tools need to develop further and be widely adopted.
Microsoft joins in
Microsoft has announced that it will contribute to Apache Ossie. That is good news for the development of the standard. The more vendors participate, the greater the chance that organisations will be able to exchange the definitions they have built between platforms. Microsoft's contribution focuses on the semantic models in Power BI and Fabric. An open-source converter can convert models to Ossie and back. In principle, this lets organisations take their model with them and reuse it elsewhere. Microsoft explains its contribution here.
The reports themselves are outside the scope of this contribution. Charts, page layouts and interactions are therefore not converted automatically. Anyone switching to another dashboard tool will still need to rebuild those reports for now.
Exchanging a model also does not guarantee that everything will work the same way on another platform. Power BI uses DAX for calculations, for example. This has now (unfortunately) been accepted by Ossie. Another platform must also be able to execute those calculations to produce the same results — and that is not yet the case. Settings such as access permissions can also be stored in the model without another tool understanding or applying them. Exporting a model is therefore a step forward, but always check which components the receiving tool actually supports. The converter documentation describes the current capabilities and limitations.
If you want to combine DataOwn with Power BI, you can connect Power BI to the transformed tables in DataOwn. You then define the relationships and metrics in Power BI. The new Ossie converter does not currently change that workflow.
Open standards make switching more feasible
DataOwn uses dbt to transform data. dbt is open source and now also supports defining and exchanging semantic models in Ossie format. This aligns with our choice of open tools and standards. The exact capabilities depend on the versions you use and the support offered by the BI tool you connect to it. More about Ossie in dbt.
The development of Apache Ossie is therefore encouraging. Organisations spend years investing in reliable metrics, clear definitions and useful reports. As much of that knowledge and logic as possible should be preserved when they switch applications or vendors.
Ossie can help increase that freedom of choice. The next step is for more tools to support the standard and for definitions, calculations and permissions to become readily transferable in practice. Until then, building with open tools and standards helps keep a switch to another vendor feasible.