Skip to main content

Performance

The Performance tab provides detailed insights into query and transaction performance. Use it to identify slow queries, analyze execution patterns, and optimize your database workload.

Performance

Sections​

The Performance tab is organized into three sections:

  1. Overview — Key performance visualizations
  2. Query Analytics — Detailed query analysis table
  3. Transactions — Transaction throughput and health

Overview Section​

The Overview section provides immediate visibility into query performance and distribution.

Average Response Time​

A time-series chart showing query response times over the selected period.

Average Response Time

What it shows:

  • Average query execution time
  • Trends over time
  • Response time variations

How to interpret:

  • Flat line at low values = healthy, consistent performance
  • Spikes = potential issues (locks, resource contention, complex queries)
  • Gradual increase = possible index degradation or data growth issues

When to investigate:

  • Response times exceeding baseline by > 20%
  • Sudden spikes correlating with errors
  • Sustained elevated response times

Query Distribution​

A pie chart showing the breakdown of query types.

Query Distribution

What it shows:

  • Proportion of SELECT, INSERT, UPDATE, DELETE operations
  • Overall query mix

How to interpret:

  • Most OLTP systems are read-heavy (SELECT > 70%)
  • Write-heavy patterns may indicate bulk operations or data ingestion
  • Unexpected changes may indicate application behavior changes

When to investigate:

  • Sudden shift in query type distribution
  • Unexpected write-heavy patterns
  • High proportion of DELETE operations (may indicate cleanup jobs)

Top Consuming Queries​

A heatmap visualization showing which queries consume the most resources.

Top Consuming Queries

What it shows:

  • Query execution time distribution
  • Resource-intensive queries highlighted
  • Time-based patterns

How to interpret:

  • Darker cells = higher resource consumption
  • Horizontal patterns = consistently expensive queries
  • Vertical patterns = time-based load spikes

How to use it:

  • Identify your most expensive queries
  • Spot queries that need optimization
  • Correlate with deployment or traffic patterns

Query Analytics Section​

A detailed table showing performance metrics for individual queries. This section is collapsed by default — click to expand.

Query Analytics

Table Columns​

ColumnDescription
QueryThe SQL query text (normalized)
CallsNumber of times the query was executed
Total TimeCumulative execution time
Mean TimeAverage execution time per call
Min TimeFastest execution
Max TimeSlowest execution
RowsTotal rows returned/affected
Shared Blks HitBuffer cache hits
Shared Blks ReadDisk reads required

How to Use​

Find slow queries:

  1. Sort by "Mean Time" descending
  2. Identify queries with high average execution time
  3. Focus optimization efforts on high-call-count slow queries

Find resource-intensive queries:

  1. Sort by "Total Time" descending
  2. These queries consume the most cumulative resources
  3. Even fast queries with high call counts can dominate

Identify caching issues:

  1. Compare "Shared Blks Hit" vs "Shared Blks Read"
  2. High read ratio = poor cache utilization
  3. Consider index improvements or memory tuning

Transactions Section​

The Transactions section shows throughput and transaction health. This section is collapsed by default — click to expand.

Transactions

Transactions Per Second (TPS)​

What it shows:

  • Number of committed transactions per second
  • Throughput trends over time

How to interpret:

  • Stable TPS = consistent workload
  • Drops in TPS = potential bottlenecks or issues
  • Spikes = batch operations or traffic surges

Benchmarking:

  • Establish baseline TPS for your workload
  • Set alerts for significant deviations
  • Use for capacity planning

Commit/Rollback Ratio​

What it shows:

  • Ratio of successful commits to rollbacks
  • Transaction success rate over time

Healthy range: > 99% commits for most workloads.

When to investigate:

  • Rollback rate exceeding 1%
  • Sudden increase in rollbacks
  • Correlation with application errors

Common causes of high rollbacks:

  • Application errors
  • Constraint violations
  • Deadlocks
  • Lock timeouts

Use Cases​

Query Optimization Workflow​

  1. Open the Performance section
  2. Identify expensive queries in the heatmap
  3. Expand Query Analytics for details
  4. Sort by "Total Time" to find biggest impact queries
  5. Analyze query patterns (missing indexes, full table scans)
  6. Implement optimizations
  7. Monitor response time improvements

Post-Deployment Performance Check​

After deploying application changes:

  1. Compare Average Response Time before/after
  2. Check Query Distribution for unexpected changes
  3. Review Top Consuming Queries for new expensive queries
  4. Verify TPS hasn't degraded
  5. Confirm Commit/Rollback Ratio is healthy

Performance Baseline Establishment​

To establish performance baselines:

  1. Monitor during typical load periods
  2. Record normal Average Response Time range
  3. Document typical TPS patterns
  4. Note expected Query Distribution
  5. Use these as reference for anomaly detection

Troubleshooting Slow Application​

When users report slow application performance:

  1. Check Average Response Time for database-level latency
  2. Look for spikes correlating with reported issues
  3. Identify slow queries in Query Analytics
  4. Check Commit/Rollback Ratio for transaction failures
  5. Review TPS for throughput bottlenecks

The Performance section uses these metrics from the Metrics Reference:

PanelPrimary Metrics
Average Response Timepg_statement_stats.avg_time_ms
Query Distributionpg_statement_stats.calls
Top Consuming Queriespg_statement_stats.total_time_ms, pg_statement_stats.calls
Query Analyticspg_statement_stats.*
TPSpg_database_stats.xact_commit
Commit/Rollbackpg_database_stats.xact_commit, pg_database_stats.xact_rollback
Was this page helpful?