Belimo: Modern data platform for smart analyses

Belimo is processing increasingly large amounts of data from connected devices. To remain powerful and efficient in the future, the company has developed a new data platform in collaboration with Ergon. The architecture separates computing and storage, uses Databricks and provides a stable foundation for data-driven innovation.

Growing data volumes and increasing requirements

Belimo's devices continuously deliver new data, causing the volume of data in the cloud to grow rapidly. The previous infrastructure was based on a time series database that had reached its performance limits. Data had to be stored on fast but expensive SSDs. In addition, the existing architecture made analysis difficult and caused high operating costs. Belimo was therefore looking for a solution that would remain economical and powerful even with further growth, while also providing modern analytics features.

Scalable architecture with a sophisticated data model

Ergon implemented a platform based on the separation of compute and storage. The data is stored in blob storage and processed via Databricks. Ergon developed a generic data model for the efficient handling of a wide variety of device data. First, all device data is stored in JSON format in a common column without Databricks needing to know the exact structure. Additional metadata ensures a simple and flexible schema. Frequently used data points are also extracted into separate columns. This allows users to quickly access relevant information while responding flexibly to new requirements. The architecture is easy to maintain, cost-efficient and expandable at any time.

Intelligent resource setup for low costs

A major cost driver in Databricks are the SQL warehouses used for queries. Belimo operates three staging environments and one productive instance. To reduce costs, Ergon developed a setup with only two workspaces. While the productive environment uses its own resources, the three staging instances share a common infrastructure. Ergon mirrored the query traffic from the live system during development. This allowed the required resources to be realistically estimated, queries to be optimised and bottlenecks to be identified at an early stage.

Future-proof foundation for data-driven services

The new platform enables Belimo to analyse large amounts of data efficiently and respond flexibly to new requirements. At the same time, it creates the basis for data-driven applications such as predictive maintenance or new IoT services.

"Thanks to the new data platform, we can now perform analyses that used to be cumbersome in no time at all and, in the medium term, at significantly reduced costs," says Philipp Stäheli, Lead Domain Software Storage Platform at Belimo.

Ergon contributed its extensive expertise in data engineering, cloud architecture and DataOps to the project. The team designed the overall architecture and data models, defined a flexible authorisation management system and established robust governance. The result is a powerful solution that meets current requirements and is ideally equipped for future innovations.