How Zalando innovates their Fast-Serving layer by migrating to Amazon Redshift

This article details how Zalando migrated its fast-serving data warehouse layer to Amazon Redshift, achieving significant performance and cost improvements.

• Zalando manages 20+ petabytes across 5,000 data products serving 6,000 monthly users
• Legacy system was 80% underutilized with $30,000+ monthly slack costs and concurrency limitations
• Redshift delivered 3-5x faster query execution and 86% of queries ran faster
• Multi-warehouse architecture separates producers from consumers using data sharing
• Three-stage migration: data replication, workload migration, finalization and decommissioning
• October 2024 switch moved 80% of analytics reporting to Redshift successfully
• Monday morning peak load time reduced from 130 to 52 minutes; 19,000 hours saved monthly
• Report timeouts virtually eliminated; Cyber Week 2024 performed exceptionally well
• Key challenges: automatic optimization awareness across warehouses, CTE recomputation, VARCHAR oversizing
• Future plans include serverless topology, AWS Lake Formation governance, and FinOps culture

Zalando's Redshift migration transformed its data platform with improved performance, stability, cost efficiency, and data consistency while maintaining operational simplicity.