Salesforce Decode
Salesforcedecode
Back to questions
Data CloudArchitect (Hardest)zero-copysnowflakeperformancefederation

Tune Zero Copy federation queries for Snowflake warehouse performance

Real World Scenario

Analysts run segment previews against Zero Copy linked tables causing Snowflake warehouse credit spikes and timeouts when joins span CRM connector data and warehouse fact tables.

Expected Answer

• Push filters to largest selective predicates early—avoid full table scans on linked objects • Materialize frequently joined warehouse aggregates as native Data Cloud insights instead of repeated federation • Align warehouse size and auto-suspend policies with segment preview usage patterns • Cache segment counts where platform supports it; warn users on expensive exploratory queries • Educate marketers on preview vs production refresh cost differences • Monitor federation query duration and credit consumption per business unit for chargeback • Evaluate hybrid model: Zero Copy for raw explore, copied DLOs for production segments

Follow-Up Questions & Answers

Click to expand — each follow-up includes a direct, interview-ready answer

Direct answer: Push filters to largest selective predicates early—avoid full table scans on linked objects Also consider: Materialize frequently joined warehouse aggregates as native Data Cloud insights instead of repeated federation In practice: Align warehouse size and auto-suspend policies with segment preview usage patterns Document the decision in an ADR and align with enterprise standards.

Architect Perspective

Zero Copy shifts compute to your warehouse bill. Architects set query guardrails and materialization strategy—not unlimited federated segment building.