Compute and Storage Separation in Cloud

🧠 Core Concept

• Compute and storage separation is a modern cloud architecture where compute and storage layers are decoupled.

• This enables independent scaling, centralized storage, and flexible resource allocation—ideal for cloud OLAP systems.

⚙️ Key Features

• Storage is centralized and accessible by all compute nodes.

• Compute resources scale independently, allowing cost optimization.

• Billing is split: storage-based and compute-based pricing.

🚀 Real-World Examples

• Snowflake

• Google BigQuery

• Amazon Redshift Spectrum

• ClickHouse Cloud

🐢 Trade-offs

• Performance may never match tightly coupled OLAP systems.

• Experts like Yury Izrailevsky and Jordan Tigani note that decoupling introduces latency and complexity.

🔍 Notable Mentions

• StarRocks and StarTree use clever format conversions (e.g., Iceberg to native) to mitigate performance hits.

• The model contrasts with shared-nothing architecture, where compute and storage are tightly bound.