snowflake
/snowflake
Wspiera pracę z platformą Snowflake: magazyny wirtualne, Snowpipe, klonowanie zero-copy, time travel i Streamlit, przy budowie hurtowni danych i analityce big data.
- Autor
- G1Joshi ↗
- Repozytorium
- Agent-Skills ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 01.10.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 1
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 1
Brak sygnałów. Repozytorium zmienione 01.10.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Wspiera pracę z platformą Snowflake: magazyny wirtualne, Snowpipe, klonowanie zero-copy, time travel i Streamlit, przy budowie hurtowni danych i analityce big data.
Oryginalny opis (EN)
Expert Snowflake cloud data warehouse assistance covering Virtual Warehouses, Snowpipe, zero-copy cloning, time travel, and Streamlit. Use when building enterprise analytical data warehouses or running big data analytics.
Jak działa
- When to Use
- Quick Start
- Core Concepts
- Common Patterns
- Best Practices
- Troubleshooting
- References
Podgląd SKILL.md
---
name: snowflake
description: Expert Snowflake cloud data warehouse assistance covering Virtual Warehouses, Snowpipe, zero-copy cloning, time travel, and Streamlit. Use when building enterprise analytical data warehouses or running big data analytics.
---
# Snowflake
Snowflake is a cloud-native data warehouse. It separates compute ("Virtual Warehouses") from storage, allowing them to scale independently.
## When to Use
- **Cloud Data Warehousing & Data Lakes**: Consolidating enterprise data across AWS, Azure, and GCP into a single analytical platform.
- **Separation of Compute and Storage (Virtual Warehouses)**: Scaling compute clusters up, down, or suspending instantly without copying data.
- **Zero-Copy Data Cloning & Time Travel**: Cloning multi-terabyte production databases in seconds for dev/test without storage cost duplication.
- **Secure Cross-Company Data Sharing**: Sharing live, governed datasets with external partners without ETL replication.
## Quick Start
```sql
-- Create warehouse (Compute)
CREATE WAREHOUSE my_wh WITH WAREHOUSE_SIZE = 'X-SMALL';
-- Query JSON directly (Variant type)
SELECT src:sales.order_id::integer
FROM raw_data;
```
## Core Concepts
### Multi-Cluster Shared Data Architecture
Decouples centralized cloud storage from independent, autoscaling virtual compute warehouses:
```text
[ Central Cloud Storage (AWS S3 / Azure Blob / GCS) ]
├── [ Virtual Warehouse: ETL (Size: X-Large) ]
├── [ Virtual Warehouse: BI Reports (Size: Medium, Autoscale) ]
└── [ Virtual Warehouse: Data Science (Size: Large) ]
```
### Time Travel & Zero-Copy Cloning
Restores historical data and creates instant instant clones without data duplication:
```sql
-- Query data as it existed 2 hours ago
SELECT * FROM analytics.orders
AT(OFFSET => -60*120)
WHERE status = 'FAILED';
-- Instant zero-copy clone for testing (takes 2 seconds, zero storage added)
CREATE OR REPLACE DATABASE dev_clone_2026 CLONE production;
```
### Native Semi-Structured VARIANT Querying
Ingests and queries nested JSON, Avro, and Parquet data directly:
```sql
SELECT
raw_payload:user_id::string AS user_id,
raw_payload:event_name::string AS event_name,
raw_payload:metadata.device.os::string AS os_system
FROM raw_logs.event_stream
WHERE raw_payload:event_name = 'checkout_completed';
```
## Common Patterns
### Zero-Copy Cloning with Time Travel Recovery
**Problem**: Creating staging replicas of terabyte datasets for testing consumes massive storage and time.
**Solution**:
Use metadata-only Zero-Copy Cloning with Time Travel:
```sql
-- Instantly clone production database without duplicating underlying micro-partitions
CREATE OR REPLACE DATABASE dev_db CLONE prod_db;
-- Recover accidentally dropped or updated table state from 2 hours ago
CREATE OR REPLACE TABLE orders_restored CLONE prod_db.public.orders
AT (OFFSET => -60*120);
```
## Best Practices
**Do**:
- Enable Auto-Suspend and Auto-Resume: Set warehouses to auto-suspend after 60 seconds of inactivity (`AUTO_SUSPEND = 60`) to stop billing.
- Cluster by Primary Query Dimensions: Apply cluster keys (`CLUSTER BY (event_date, customer_id)`) on multi-terabyte tables to optimize micro-partition pruning.
- Leverage Transient Tables for Staging: Use `CREATE TRANSIENT TABLE` for intermediate ETL steps to eliminate Time Travel storage costs.
- Use Dynamic Data Masking: Protect sensitive PII columns using role-based masking policies.
**Don't**:
- Leave oversized virtual warehouses running 24/7: Size warehouses appropriately; scale down when batch jobs complete.
- Use Snowflake for single-row transactional OLTP: High latency per single insert makes Snowflake unsuitable for low-latency operational backends.
- Ignore micro-partition pruning: Always filter queries by date or clustered keys to avoid full table scans.
## Troubleshooting
| Error | Cause | Solution |
| :-------------------------------------------------- | :----------------------------------------------------------- | :-------------------------------------------------------------------- |
| `Warehouse '...' has exceeded maximum credit limit` | Long-running queries keeping large virtual warehouse active. | Configure `AUTO_SUSPEND = 60` and `AUTO_RESUME = TRUE` on warehouses. |
| `Query compilation error: ambiguous column name` | Joins referencing shared column names without table aliases. | Qualify all column references with explicit table aliases. |
| `Snowpipe load error: File size exceeds maximum` | Staged file too large for streaming ingestion. | Split input files into optimal 100MB-250MB compressed chunks. |
## References
- [Snowflake Documentation](https://docs.snowflake.com/)
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · G1Joshi/Agent-Skills | 100 | MIT |
| adobe/skills | 100 | Apache-2.0 |
Migawka bazy z 13.11.2026 · wersja paczki 8519d9f6c99d (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 100snowflake-data-engineering
Zbiera dobre praktyki dla Snowflake SQL, danych półstrukturalnych i potoków z Dynamic Tables, Streams, Tasks i Snowpipe, w tym dostrajanie wydajności i kosztów, Time Travel, klonowanie, RBAC i tabele Iceberg. - 100snowflake-expert
Wspiera pracę z hurtownią danych Snowflake: architekturę i magazyny wirtualne, obiekty bazy, ładowanie danych i stage'e, udostępnianie danych, streamy, taski i optymalizację SQL. - 100snowflake-data-cloud-architecture
Projektuje w Snowflake współdzielone dane z wieloklastrowymi hurtowniami wirtualnymi i klonowaniem bez kopiowania danych. - 100snowflake-development
Optymalizuje zapytania Snowflake, analizuje wydajność SQL, dobiera rozmiar magazynów, przegląda modele danych i diagnozuje problemy z kosztami. - 90snowflake-native-pipelines-and-governance
Prowadzi agentów przez natywne przepływy Snowflake: pipeline'y ze Streams, Tasks, Dynamic Tables, Snowpipe i Snowpark oraz maskowanie, dostęp do wierszy i bezpieczne udostępnianie. - 95setup-warehouse-snowflake
Konfiguruje Snowflake jako hurtownię danych dla Confidence, gdy użytkownik wybrał tę platformę.