.webp){kind=spacer}
When your business is helping millions explore the world, your data infrastructure shouldn’t hold you back.
That was the realization for GetYourGuide, one of the world’s leading travel experience platforms. As data usage and business intelligence workloads continued to grow, their existing data stack, built on Snowflake, was increasingly challenged by performance and cost considerations.
This wasn’t a failure of Snowflake, but a reflection of how GetYourGuide’s data needs had evolved. To support a more centralized and scalable data platform, GetYourGuide decided to evaluate alternatives and ultimately turned to Databricks, partnering with Datapao to support both the proof of concept and the migration.
Why migrate from Snowflake to Databricks?
Snowflake had long supported GetYourGuide’s business intelligence needs, with Looker serving as the primary BI tool. However, the company’s broader data platform already relied on Databricks as the source of truth for data processing, transformations, and analytics.
Maintaining Snowflake as a separate query layer introduced additional data movement and operational overhead. To simplify the architecture and reduce costs, GetYourGuide evaluated whether Databricks could also support their BI workloads at the required scale and performance.
Key challenges and evaluation criteria included:
- Query performance under high concurrency
- Operating costs of the query engine
- Architectural complexity caused by maintaining separate systems for processing and BI
To meet these requirements, GetYourGuide needed more than incremental optimization. They needed a platform that could unify data processing and analytics while maintaining strong BI performance. Databricks, built on the Lakehouse architecture, offered a compelling alternative.
Making migration look easy (even when it isn’t)
A Snowflake to Databricks migration is never “just a switch.” Done poorly, it risks disrupting reporting and slowing down business users.
Datapao’s role focused on supporting a controlled, performance-focused migration, helping GetYourGuide validate assumptions during the proof of concept and optimize the platform during the migration phase. The goal was to ensure a smooth transition that preserved report correctness and performance.
The approach included:
- A proof of concept to benchmark Databricks against Snowflake for BI workloads
- Configuration-driven table optimization and migration from Parquet to Delta
- Support for optimizing queries across more than 1,200 tables
- BI integration and optimization by connecting Looker to Databricks Serverless SQL Warehouses
- Automated performance and correctness validation using the Looker SDK and APIs
Under the hood: real engineering, real results
Behind the scenes, the migration relied on iterative engineering work and systematic optimization.
Performance optimization
Rather than relying on a single lift-and-shift approach, performance was improved through repeated optimization cycles. These included SQL query adjustments, Delta table configuration tuning, and SQL warehouse sizing experiments. Some workloads performed better immediately, while others required additional optimization to reach or exceed Snowflake performance.
Fully automated Delta Lake maintenance
To support long-term performance and manageability, the migration introduced configuration-driven processes for loading and maintaining Delta tables. These processes simplified table management, and ensured that tables remained optimized as data volumes grew.
From proof of concept to production
The proof-of-concept validated that Databricks could meet GetYourGuide’s BI performance and cost targets. This enabled the team to move forward with a full migration using an incremental rollout strategy designed to minimize risk.
After a complete optimization cycle:
- 72% of Looker queries ran under 10 seconds
- The migration was completed in approximately 4.5 months, faster than typical industry benchmarks for similar workloads
- No significant reporting issues were observed during the rollout
As a result, analysts were able to explore data efficiently, and engineers could operate on a more centralized and simplified data platform.
Not just a migration, a platform evolution
This Snowflake to Databricks migration was more than a backend change. It represented a shift toward a more unified data platform, where data processing and business intelligence operate on the same foundation.
With centralized data processing, reusable migration and validation tooling, and automated table management in place, GetYourGuide’s teams are now positioned to continue optimizing performance and cost over time, building on the foundation established during the migration.
“The collaboration with Datapao ran smoothly, with quick turnarounds. They helped us assess the value and complexity of the migration, and we delivered on time.”
Thiago Rigo, Engineering Manager, GetYourGuide
Thinking about your own Snowflake to Databricks migration?
Whether you’re looking to reduce costs, simplify your architecture, or consolidate BI and data processing on a single platform, you don’t have to go it alone. Datapao has supported complex Snowflake to Databricks migrations across BI, data engineering, and analytics environments.
Let us help you turn friction into flow.
Book a migration consultation
Special thanks to the GetYourGuide data team for their trust, partnership, and insights throughout the engagement.
.webp)
.webp)
.webp)
