Distribution & Logistics
Critical pipeline 8x faster and 20% lower operating costs
A new foundation on Azure and a first-of-its-kind data streaming architecture for Unilever Brasil's largest digital distributor.
Case metrics
8x
Faster: critical pipeline from 4h to 30min
-20%
Operating costs
Real time
A streaming architecture that did not exist before
The project story
Challenge
e-coa is Unilever Brasil's largest digital distributor and serves independent retail.
At the start of the project, we identified improvements needed across its technology landscape, from how data was consumed to the security and data delivery layers.
Approach
Moving to Azure created a scalable, resilient foundation, with near-zero downtime and failover protection for the main cloud services.
On top of it, we built a reference architecture based on Data Lakehouse, with self-service BI and data governance.
For real-time logistics, we created a data streaming architecture with Confluent and Databricks: TableFlow for the pipelines, Flink SQL for the business rules, and Structured Streaming for continuous analytical consumption.
Result
The critical pipeline became 8x faster, from 4 hours to 30 minutes, and operating costs dropped 20% with the optimization of logistics processes.
Business areas get insights faster, and 100% of the company was impacted by the Power BI dashboards.
Project highlights
- Azure foundation with near-zero downtime
- Data Lakehouse reference architecture
- Self-service BI with data governance
- Real-time pipelines with TableFlow
- Business rules and transformations in Flink SQL
- Continuous analytical consumption with Databricks Structured Streaming
Technologies
- Confluent
- Databricks
- Azure
Customer testimonial
Without a doubt, we took a considerable leap in terms of technology. We started using the leading market tools, which has helped us be more competitive by delivering insights faster to the business areas, and also adopting governance policies for processes and data quality, with the processes increasingly in our hands.
Other cases
What does it cost to wait hours for a pipeline that could take minutes?
Talk to Datatchê to rebuild your data foundation and deliver insights to the business faster.
