Export dashboard as PDF for board reporting
Databricks reduces the gap between raw data and actionable business insight from days or weeks to minutes by combining serverless SQL warehouses, collaborative notebooks, and AI-powered assistants in a single workspace. Business teams can query governed data directly without waiting for engineering handoffs.
Who this is for:
Part of the How Databricks Can Help Your Business section of the Databricks tutorial series.
Architecture / Concept Overview: Export dashboard as PDF for board reporting
The traditional insight workflow involves filing a ticket, waiting for engineering to prepare data, then building a report. Databricks collapses this by giving business teams direct, governed access to curated datasets via SQL and dashboards.
*Figure 1 — From business question to dashboard in minutes using serverless SQL.*
*Figure 2 — Traditional ticket-based workflow vs self-service analytics on Databricks.*
*Figure 3 — Proactive alerting ensures business teams learn about issues before they escalate.*
Key Terms
Prerequisites and Setup
- Databricks workspace with a serverless SQL warehouse enabled
- Unity Catalog configured with at least one gold-layer schema
- Business user accounts provisioned with SELECT access to relevant schemas
- A notification channel (email or Slack) configured for alerts
Step-by-Step Implementation
Configuration Reference
| Parameter | Description | Recommended Value |
|---|---|---|
| Warehouse type | Serverless vs provisioned | Serverless for ad-hoc |
| Auto-stop | Minutes before idle shutdown | 10 minutes |
| Query result cache | Cache duration for repeated queries | 1 hour |
| Dashboard refresh | Auto-refresh interval | Every 15-60 minutes |
| Alert evaluation | How often alert conditions are checked | Match query schedule |
| Concurrency limit | Max concurrent queries per warehouse | 20-50 for shared teams |
Monitoring, Cost, and Security Considerations
Monitoring
Track query performance via the Query History interface. Monitor warehouse utilisation and queue times. Set up alerts for queries exceeding latency thresholds so engineers can optimise underlying tables.
Cost Optimisation
Use serverless warehouses to avoid paying for idle compute. Enable query result caching to reduce redundant execution. Right-size warehouses based on actual concurrency — most analyst teams work well with a medium warehouse.
Security and Governance
Business users should only access gold-layer tables with pre-applied transformations and masking. Use dynamic views to implement row-level security (e.g., regional sales teams see only their region). Audit all query activity through system tables.
Common Pitfalls and Recommended Patterns
- Giving analysts access to bronze/silver tables — they encounter raw, unfamiliar schemas and generate incorrect reports
- Over-provisioning warehouse size for low-concurrency teams — start small and scale on demand
- Building dashboards without query parameters — leads to proliferation of nearly identical dashboards
- Skipping result caching — identical queries re-execute unnecessarily, increasing cost
- Not training business users on the AI assistant — they default to filing tickets instead of self-serving
- Creating alerts without clear ownership — notifications without assigned responders get ignored
Frequently Asked Questions
How fast can a business user get their first answer?
With serverless warehouses and pre-curated gold tables, a user can run their first query within seconds of logging in. No cluster startup delay.
Do business users need to know SQL?
Basic SQL helps, but the AI assistant can generate queries from natural language descriptions. Users can also interact purely through pre-built dashboards.
Can we connect our existing BI tools?
Yes. SQL warehouses expose standard ODBC/JDBC connectivity. Tableau, Power BI, Looker, and other BI tools connect natively with optimised drivers.
How do we handle ad-hoc requests that need new data?
Self-service for existing gold tables is instant. If a request requires new data sources or transformations, engineers add them to the pipeline and publish new tables to the catalog — but the analyst never waits for a one-off data dump.
What happens when many analysts query simultaneously?
Serverless SQL warehouses auto-scale to handle concurrent load. Queries queue briefly during extreme spikes, but the system scales within seconds.