Back to all Blogs
1 Sep 2025 Data Platforms

The Ultimate Guide: How to Optimise Snowflake Costs on Azure & AWS

Learn 10 proven strategies to optimise Snowflake costs across Azure and AWS, balancing performance, scalability, and cost-efficiency.

The Ultimate Guide: How to Optimise Snowflake Costs on Azure & AWS

Introduction: The Double-Edged Sword of Cloud Data Warehousing

Snowflake has revolutionised the world of cloud data warehousing with its separation of compute and storage, scalability, and pay-as-you-go pricing model. Running on major clouds like Azure and AWS, it allows enterprises to process massive workloads without upfront infrastructure costs.

But here’s the catch: that same flexibility can make costs unpredictable and hard to control.

  • For executives, this means budget uncertainty.
  • For technical leads, it means tighter governance and accountability.

This guide covers 10 proven strategies to optimise Snowflake costs across Azure and AWS—balancing performance, scalability, and cost-efficiency.

Understand Snowflake’s Pricing Model

Snowflake charges you in three key areas:

  • Compute (Virtual Warehouses) – Billed per-second (rounded to the nearest minute). The biggest cost driver, typically 70–80% of spend.
  • Storage – Costs for compressed data, Time Travel, and Fail-safe.
  • Data Transfer – Costs when data moves across regions/clouds.

Manager’s Tip: Request a monthly breakdown by category—this highlights optimisation opportunities quickly.

Right-Size Your Virtual Warehouses

  • Start small (XS/S warehouses) and scale as workloads grow.
  • Use multi-cluster warehouses only if concurrency is a proven issue.
  • Regularly analyse query performance and usage history.

Technical Tip: Use WAREHOUSE_METERING_HISTORY to spot underutilised warehouses.

Stop Paying for Idle Compute

  • Enable auto-suspend (2–5 mins prod, 1 min dev/test).
  • Turn on auto-resume for seamless performance.
  • Schedule shutdowns for dev/test on nights & weekends.

Many teams cut 30–40% off monthly costs with this single setting.

Control Spend with Resource Monitors

  • Set credit quotas by warehouse, team, or department.
  • Add alerts at 75% and 90% usage thresholds.
  • Optionally suspend warehouses automatically when limits are hit.

Manager’s View: Prevents cost overruns before they happen.

Optimise Storage Strategy

  • Purge unused/stale tables.
  • Use transient/temporary tables for staging data.
  • Reduce Time Travel for non-critical tables from 7 days → 1 day.

Cloud Note: Extra retention + replication = bigger AWS/Azure bills.

Avoid Costly Data Transfers

  • Keep compute + storage in the same region.
  • Replicate only when compliance demands it.
  • Use Snowflake Secure Data Sharing instead of exporting large files.

Golden Rule: Bring queries to the data, not data to the queries.

Write Efficient Queries

  • Avoid SELECT *—fetch only required columns.
  • Add clustering keys on large tables.
  • Use result caching & materialised views for recurring queries.
  • Check QUERY_HISTORY for runaway queries.

Training analysts = long-term cost savings.

Make Cost Visible Across Teams

  • Query Account Usage schema (WAREHOUSE_METERING_HISTORY, STORAGE_USAGE, etc.).
  • Build dashboards in Power BI/Tableau.
  • Integrate with Azure Monitor or AWS CloudWatch.

Manager’s Insight: Visibility changes behaviour—teams become self-aware.

Negotiate Better Pricing

  • Explore enterprise contracts with upfront commitments.
  • Use Azure/AWS enterprise agreements for bundled discounts.
  • Review usage & renegotiate annually.

Savings: 10–20% without any technical changes.

Build a FinOps Culture

  • Tag workloads for chargeback/showback.
  • Run monthly reviews for anomalies.
  • Educate teams on best practices.
  • Treat credits as currency, not infinite compute.

Executive Takeaway: FinOps empowers teams to innovate without overspending.

Bonus: Azure vs AWS Considerations

  • Azure: Strong ties with Power BI & Synapse; monitor egress with non-Azure services.
  • AWS: Ecosystem rich in S3, Glue, Redshift integrations; higher risk of cross-region transfer costs.

Strategy is the same: keep compute local, monitor aggressively, optimise continuously.

Conclusion: Smarter Spending, Not Just Cost Cutting

  • Cut wasted compute
  • Keep storage lean
  • Prevent surprise bills
  • Build a culture of cost ownership

The goal isn’t to spend less—it’s to spend smarter, while keeping performance and agility intact.

The Ultimate Guide: How to Optimise Snowflake Costs on Azure & AWS | Winsys Blog