Fail-Safe Storage: The Cost You Can't Disable But Can Minimize

Fail-Safe Storage: The Cost You Can't Disable (But Can Minimize)

Fail-safe adds 7 days of storage costs you can't opt out of. Learn strategies to minimize its impact on your overall storage bill.

How Fail-Safe Works

• Automatic 7-day recovery period after Time Travel expires
• Cannot be disabled or configured
• Only accessible by Snowflake support
• Costs the same as regular storage

Storage Timeline

Day 0: Data written
Day 0-N: Time Travel period (0-90 days, configurable)
Day N to N+7: Fail-safe period (7 days, non-configurable)
Day N+8: Data permanently deleted

Cost Impact Example

Table size: 10TB, Daily updates: 1TB, Time Travel: 1 day

Total storage = 10TB (current) + 1TB (Time Travel) + 7TB (Fail-safe) = 18TB

You're paying for 80% more storage than the actual table size!

Strategies to Minimize Fail-Safe Costs

1. Reduce Time Travel Retention

ALTER TABLE staging_data SET DATA_RETENTION_TIME_IN_DAYS = 0;

2. Use Transient Tables

CREATE TRANSIENT TABLE staging_data (
id INTEGER,
data VARCHAR
);

Transient tables have no Fail-safe period (but also no data recovery beyond Time Travel).

3. Optimize Update Patterns

-- GOOD: Incremental merge
MERGE INTO production_data t
USING source_data s ON t.id = s.id
WHEN MATCHED THEN UPDATE SET t.value = s.value
WHEN NOT MATCHED THEN INSERT VALUES (s.id, s.value);

When to Use Transient vs Permanent Tables

Transient (No Fail-safe):

• Staging/ETL intermediate tables
• Development and testing
• Data that can be easily recreated

Permanent (With Fail-safe):

• Production data
• Financial/compliance data
• Critical business data

Real-World Example

A data warehouse had 100TB of production tables with high daily churn. Fail-safe was costing $2,000/month. By converting 40TB of staging tables to transient (saved $800/month), reducing Time Travel from 7 to 1 day on 30TB of tables (saved $400/month), and optimizing ETL to use incremental updates (saved $300/month), they achieved $1,500/month savings (75% reduction).

Key Takeaways

• Fail-safe adds 7 days of unavoidable storage costs
• Use transient tables for staging and ETL
• Reduce Time Travel to minimize total retention period
• Incremental updates generate less Fail-safe data