Exam Focus · Part 2
Read the symptom first, then pick the cheapest fix that works: size, clusters, a pruning service or the SQL.
Know these cold
| Topic | Remember |
|---|---|
| Credit rates | X-Small (Gen1) = 1 credit/ |
| Scale up vs out | Spilling or one heavy query: bigger size. Queuing: more clusters |
| Multi-cluster | Enterprise. Maximized: min = max. Auto-scale: min < max. Clusters all one size |
| Max clusters | 10 in Snowsight. SQL: up to 300 (XS–M), 10 at 4XL+ |
| Scaling policy | Auto-scale only. Standard: start on queuing. Economy: needs 6 minutes of work |
| Resource monitor | ACCOUNTADMIN creates it. One per warehouse, plus one account-level |
| Monitor actions | SUSPEND lets running queries finish. SUSPEND_ |
| Budget | Sees serverless and AI spend too. No built-in suspend; a custom action can |
| Full scans | Clustering: wide range filters. SOS (Enterprise+): selective lookups |
| QAS | Enterprise, per warehouse, for outlier scans. On by default for new standard warehouses |
Traps
- Myth Resizing cancels running queries.Fact They finish on the old size.
- Myth A bigger warehouse fixes an exploding join.Fact It's a missing join condition. Fix the SQL.
- Myth A resource monitor can cap Snowpipe or clustering spend.Fact Monitors see warehouse credits only.
- Myth QAS scale factor 0 turns acceleration off.Fact 0 means no upper bound.
Before you turn the page
- Recite the credit ladder from X-Small to 4X-Large.
- Give the fix for each symptom: spilling, full scan, exploding join, queuing.
- Say what SUSPEND and SUSPEND_IMMEDIATE do to running queries.

