Performance Optimization
Singdata Lakehouse provides multi-level performance optimization capabilities covering query acceleration, storage optimization, index recommendations, and problem diagnostics — applicable to different performance bottleneck scenarios.
This Section
| Page | Description |
|---|---|
| Result Cache | Query result caching: identical queries hit the cache directly without recomputation — ideal for high-frequency repeated queries |
| Compute Cluster Cache | Preload hot data to compute cluster local nodes to reduce object storage I/O latency; supports active caching (AP clusters) and passive caching (GP/AP clusters) |
| Small File Optimization | Automatically merge small files generated by high-frequency writes to reduce I/O during queries; suitable for high-frequency write scenarios like Dynamic Table refreshes |
| Recommended Sorting Columns for Tables | The system automatically analyzes query filter conditions and recommends columns suitable for Sort Keys to speed up filter queries |
| Job Profile | View job history from the past 7 days, analyze slow queries and failed jobs, and identify performance bottlenecks |
Quick Selection Guide
Query results don't change but every run recomputes from scratch → Enable Result Cache — identical SQL hits the cache directly
BI report queries are slow and data is on object storage → Use Compute Cluster Cache to actively preload hot tables to AP cluster local storage
Queries slow down after Dynamic Table or high-frequency writes → Enable Small File Optimization to merge fragmented files
Not sure which columns should be Sort Keys → Enable Auto Index Recommendations — the system suggests based on actual query patterns
Slow queries or errors with unknown causes → Open Job Profile to view execution details and error information