Why Snowflake is Slow for Transformations and My 4 Insights from dbt Coalesce 2023

When it comes to Business Intelligence (BI), Snowflake is often lauded for its speed and ‘set-it-and-forget-it’ simplicity. However, while attending dbt Coalesce in San Diego, I discovered a different narrative among Analytics Engineers. They were saying “Snowflake is slow to build my models.” You see, dbt handles most of the transformations in cloud data warehouses. How else do you do version-controlled templated SQL for your ELT (Extract, Load, Transform) pipelines in a cloud data warehouse? These folks were not using these warehouses for BI, they were using them for transformations. The usual benchmarks for <a href="https://blocksandfiles.com/2021/11/15/snowflake-rebuts-databricks-snowflake-performance-comparison/" rel="noopener ugc nofollow" target="_blank">TPC-DS</a> (TPC-DS is the industry standard benchmark for query serving of data warehouse data from the <a href="https://tpc.org/" rel="noopener ugc nofollow" target="_blank">non-profit TPC</a> founded to define database benchmarks to disseminate objective, verifiable performance data to the industry) don’t align with the real-world work carried out by Analytics Engineers. <a href="https://medium.com/dbsql-sme-engineering/why-snowflake-is-slow-for-transformations-and-my-4-insights-from-dbt-coalesce-2023-c05af48e4298">Click Here</a>