Overview

A leading global gaming company  with a data-driven culture powering player personalization, game performance analytics, and real-time operations partnered with Koantek to modernize its data platform, completing a complex migration from Snowflake to Databricks. In just four weeks, Koantek successfully transitioned over 10,000 tables, 3,000 queries, and 1,000 workflows, moving nearly 3 petabytes of data—the largest Databricks migration on record—while maintaining data integrity and performance throughout.

Key Impacts

• 3 PetaBytes of data migrated with zero downtime
• 40% Reduction in operational costs
• 20% Improved efficiency
• 48% Increase in productivity for data engineers

Challenges Faced by Customer

• Scalability Constraints: As game data volume surged, Snowflake costs and processing limitations began to impact efficiency.
• Platform Fragmentation: Data engineering, analytics, and orchestration were distributed across multiple systems, creating bottlenecks and maintenance challenges.
• Complex Data Dependencies: With over 10,000 tables and workflows built in legacy orchestration tools, migrating without disrupting operations was a high-risk undertaking.
• Unsupported SQL Functions: Certain Snowflake functions lacked direct support in Databricks, requiring manual intervention or alternatives.
• Language & Localization: Handling multilingual data, including Chinese-language datasets, added complexity to validation and transformation.

How Koantek Responded

Koantek partnered closely with the customer’s internal teams and Databricks engineers to deliver a structured, phased migration:

• Discovery & Planning: Conducted in-depth analysis of ingestion patterns, workflows, and platform dependencies.
• Data Migration: Migrated data from Snowflake to Amazon S3, then into Databricks, achieving speeds of 200TB per day.
• Schema & Query Conversion: Rebuilt schemas and transpiled 3,000+ SQL queries using Databricks’ Snowflake transpiler, supported by Koantek’s custom validation tool.
• Workflow Migration: Moved 1,000 orchestration workflows from Airflow and Dolphin (Themis 2) into Databricks Workflows, sequencing across Bronze, Silver, and Gold layers.
• Validation & Testing: Implemented automated, snapshot-based validation checks for data accuracy across environments.
• Enablement & Monitoring: Trained internal teams, deployed Unity Catalog, and enabled real-time monitoring via Databricks UI.

Koantek addressed technical gaps (e.g., unsupported SQL syntax) with custom workarounds in collaboration with Databricks Resident Solution Architect.

Outcome

• 3 Petabytes of data migrated in under four weeks — the largest data migration in Koantek’s history
• 2,456 tasks transpiled and 895 workflows migrated with zero downtime during business-critical operations
• Query performance improved, meeting  SLA benchmarks
• 40% Reduction in operational costs
• 20% Improved efficiency
• 48% Increase in productivity for data engineers
• 100% data model fidelity maintained, with full lineage and transformation tracking
• Enabled self-service orchestration and analytics for internal teams through automation and training
• Migration laid the foundation for machine learning, predictive analytics, and real-time reporting

Testimonial

"The migration exceeded our expectations in both execution and impact. Koantek's structured approach, custom tooling, and close collaboration with our teams made a complex platform transition feel seamless. This was not just a migration—it was a launchpad for our next generation of data innovation."