Complete Solutions
What is Snowflake? (key features) - ANSWER - Analytic data warehouse
SaaS offering
- No hardware/installation/patching
- No ongoing maintenance/tuning
- Can't run privately (on-prem/hosted)
Runs completely in the Cloud (AWS, Azure, GCP)
- Has its own VPC
Decoupled compute and storage (scaled compute does not need scaled storage)
Snowflake Architecture - Cloud Services - ANSWER - Command center of Snowflake
- Coordinates and ties all activities together within Snowflake
Gets provisioned by Snowflake and within cloud provider (AWS, Azure, GCP)
- Customer can't access directly or modify it
- Shared with other customers unless using VPS Edition
Snowflake Pricing (what determines unit cost, pricing model, editions features) - ANSWER - Unit
costs for Credits and data storage determined by Region (not Cloud platform)
Pricing model = on-demand or discounted upfront
Editions:
,- Enterprise (or greater) => 90 days time-travel (default is 1 day) + materialized view + multi-
cluster warehouse
- Business Critical => more security (HIPAA, SOC 1&2, PCI DSS)
- VPS Edition => own cloud service layer (not shared with accounts)
Snowflake Regions - Is multi-region account supported? - ANSWER - No multi-region accounts
supported
- Each account can only be in one region
Multiple regions supported for each cloud provider
- AWS: 11 regions (US, EU, Asia-Pacific)
- Azure: 8 regions (US, EU, Asia-Pacific)
- GCP: 3 regions (US, EU)
Options to Connect to Snowflake - ANSWER - Web-based UI
Command-Line client (SnowSQL)
ODBC, JDBC (must download driver)
Native connectors (Python, Spark, Kafka)
Third-Party connectors (e.g., Matillion)
Other (Node.js, .Net)
Connect URL Pattern - ANSWER -
https://<account_name>.<region>[.provider].snowflakecomputing.com
example for AWS:
https://pp12345.ap-southeast-2.snowflakecomputing.com
, Snowflake Architecture Overview - ANSWER - Hybrid of shared-disk db and shared-nothing db
- Uses central data repo for persisted storage - all compute nodes have data access
- Uses MPP clusters to process queries - each node stores a portion of the data locally
Snowflake Architecture - 3 main Layers - ANSWER - Storage Layer
Query Processing
Cloud Services
Snowflake Architecture - Storage Layer - ANSWER - All data stored as internal, optimized,
compressed columnar format (micro-partitions represents logical structure of table)
Can't access data directly, only through Snowflake
Snowflake Architecture - Query Processing - ANSWER - Where processing of query happens
Uses virtual warehouses - MPP compute cluster
- Each cluster is independent (doesn't share resources, with, doesn't impact other Virtual WHs)
What does the Cloud Services Layer handle? - ANSWER - Authentication
Infrastructure Management
Metadata Management