Database / Snowflake Interview Questions
What is Snowflake and how does its multi-cluster, shared-data architecture differ from traditional data warehouses?
Snowflake is a cloud-native data platform delivered as a fully managed SaaS, built on cloud object storage (S3, Azure Blob, GCS). Its defining architectural trait is complete decoupling of storage from compute, setting it apart from every traditional warehouse model.
Traditional warehouses use either shared-nothing (Teradata, Redshift) where each node owns local disks — adding compute forces proportional storage growth — or shared-disk (Oracle RAC) where a central storage bus creates a bottleneck. Snowflake's multi-cluster shared-data model stores all data in cloud object storage and lets any number of independent Virtual Warehouses read from it simultaneously with no contention. An ETL warehouse and a BI warehouse can query the same petabyte at the same time, each billed only for their own compute seconds.
| Aspect | Shared-Nothing (Redshift) | Snowflake Multi-Cluster |
|---|---|---|
| Storage/compute coupling | Tightly coupled per node | Fully decoupled |
| Multiple concurrent workloads | One cluster, resource contention | Independent warehouses, no contention |
| Scale compute alone | No — nodes carry their own storage | Yes — resize warehouse without touching data |
| Idle cost | Pay for cluster 24/7 | Suspend warehouse = zero compute billing |
More Related questions...