Transactional data lake architecture pattern for unified analytics, AI/ML, and other collaborative workloads.
Ensure flexibility in handling diverse data types by storing structured and semi-structured data within a single repository.
Iceberg Tables (private preview) create an open lakehouse in your cloud storage with the leading standard for open table formats, Apache Iceberg™.
No upfront ingest costs for onboarding existing Apache Iceberg™ datasets to Snowflake.
For both Snowflake-managed and external data lake storage, know and protect your data with Snowflake’s breadth of native data governance capabilities.
Monitor sensitive data for compliance, discovery, protection, and resource usage with Object Tagging
Enforce column-level security with Dynamic Data Masking and External Tokenization
Automate masking policy enforcement with tag-based masking
Enforce row-level security with Row Access Policies
Use built-in, ready-to-query views to perform regulatory compliance, auditing, understand lineage, usage patterns, and relationships
Directly access curated and modeled data from other internal business units, external organizations, and from over 2,000 ready-to-query datasets on the marketplace (as of July 31, 2023).
Grant revocable access to live Iceberg and Snowflake data sets, data services, and applications that remain in place.
Power highly concurrent workloads without resource contention.
Near-zero performance tradeoff between Snowflake-managed Iceberg Tables and ingested Snowflake-format tables
Consumption-based pricing ensures you’re only charged for what you use
Automatic scaling of compute resources based on workload requirements
Compute isolation minimizes resource contention
Budgets (public preview) and Resource Monitors to monitor and control spend
Ongoing performance improvements and cost optimization means your price for performance only gets better
Easily build powerful, polyglot pipelines on Snowflake with the option to integrate with external engines.
Use Snowpark to code in Python, Scala, and Java with the simplicity and security of running from a single, multi-cluster compute engine.
Retain flexibility to connect an external engine at no cost.