This report quantifies the queries per second (QPS) impact of enabling and configuring Performance Schema on ApsaraDB RDS for MySQL and PolarDB for MySQL. Use the data to weigh monitoring capability against performance overhead before adjusting your Performance Schema settings.
Key findings at concurrency 32: Disabling Performance Schema entirely (pfsoff) gives the highest QPS. Enabling full monitoring (pfson_conon_inson) reduces QPS by up to ~23% on RDS and ~24% on PolarDB in write-only workloads. Enabling the DAS Performance Insight configuration (pfson_conon_ins_sqlon) adds 13–22% overhead while providing statement-level diagnostics.
Testing scenario
The following table describes the test environment.
| Attribute | Value |
|---|---|
| Databases | ApsaraDB RDS for MySQL, PolarDB for MySQL |
| Test tool | SysBench |
| Test tables | 10 |
| Concurrency levels | 1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1,024 |
Instance and cluster specifications
| Database | Edition | Instance family | Instance type | CPU and memory | Max connections | Max IOPS |
|---|---|---|---|---|---|---|
| ApsaraDB RDS for MySQL | High-availability Edition | General-purpose | mysql.n4.large.2c | 4 cores, 16 GB | 8,000 | 2,800 |
| PolarDB for MySQL | Cluster Edition | Dedicated | polar.mysql.x4.large | 4 cores, 16 GB | 8,000 | 48,000 |
SysBench is a modular, open-source, cross-platform, multi-threaded benchmark tool for evaluating database performance under heavy load. For more information, see the SysBench documentation.
Scenarios with different parameter settings
The following scenarios are defined by combinations of the performance_schema parameter and the consumers and instrumentation configuration items. Each scenario represents a different level of monitoring capability.
For information about how to modify the performance_schema parameter, see Modify the parameters of an ApsaraDB RDS for MySQL instance.
| Scenario | Performance Schema | Consumers | Instrumentation | Monitoring capability |
|---|---|---|---|---|
| pfsoff | Disabled | — | — | None. Baseline performance. |
| pfson_conoff_insoff | Enabled | All OFF | All OFF | Minimal — Performance Schema is on but collects no data. |
| pfson_conon_insoff | Enabled | All ON | All OFF | Partial — consumer pipeline active, but no instrumentation points are tracked. |
| pfson_conon_ins_sqlon | Enabled | Selected ON (see below) | Statement instrumentation ON | Statement-level statistics. This is the configuration used by the new version of the DAS Performance Insight feature. |
| pfson_conoff_inson | Enabled | All OFF | All ON | Instrumentation active, but no consumers process the data. |
| pfson_conon_inson | Enabled | All ON | All ON | Full monitoring — all events and statistics collected. Highest overhead. |
SQL commands for each scenario
pfson_conoff_insoff
-- Disable all consumers
UPDATE performance_schema.setup_consumers SET Enabled = 'NO';
-- Disable all instrumentation
UPDATE performance_schema.setup_instruments SET Enabled = 'NO', timed = 'NO';pfson_conon_insoff
-- Enable all consumers
UPDATE performance_schema.setup_consumers SET Enabled = 'YES';
-- Disable all instrumentation
UPDATE performance_schema.setup_instruments SET Enabled = 'NO', timed = 'NO';pfson_conon_ins_sqlon
-- Enable selected consumers: global_instrumentation, thread_instrumentation,
-- events_statements_current, events_statements_history, statements_digest
UPDATE performance_schema.setup_consumers
SET enabled = IF(name IN (
'global_instrumentation',
'thread_instrumentation',
'events_statements_current',
'events_statements_history',
'statements_digest'
), 'YES', 'NO');
-- Enable statement instrumentation only
UPDATE performance_schema.setup_instruments
SET enabled = IF(name LIKE 'statement/%', 'YES', 'NO'),
timed = enabled;pfson_conoff_inson
-- Disable all consumers
UPDATE performance_schema.setup_consumers SET Enabled = 'NO';
-- Enable all instrumentation
UPDATE performance_schema.setup_instruments SET Enabled = 'YES', timed = 'YES';pfson_conon_inson
-- Enable all consumers
UPDATE performance_schema.setup_consumers SET Enabled = 'YES';
-- Enable all instrumentation
UPDATE performance_schema.setup_instruments SET Enabled = 'YES', timed = 'YES';Test results
Average QPS at concurrency level 32, measured after CPU utilization reaches 100%. The pfsoff scenario is the baseline (Performance Schema disabled). All other rows show QPS reduction relative to that baseline.
ApsaraDB RDS for MySQL
| Scenario | Read-only QPS | vs. baseline | Read/write QPS | vs. baseline | Write-only QPS | vs. baseline |
|---|---|---|---|---|---|---|
| pfsoff | 47,176 | Baseline | 37,999 | Baseline | 39,940 | Baseline |
| pfson_conoff_insoff | 41,111 | −12.86% | 33,170 | −12.71% | 32,327 | −19.06% |
| pfson_conon_insoff | 41,773 | −11.45% | 33,568 | −11.66% | 31,434 | −21.30% |
| pfson_conon_ins_sqlon | 39,889 | −15.44% | 33,056 | −13.01% | 31,865 | −20.22% |
| pfson_conoff_inson | 39,344 | −16.61% | 32,698 | −13.95% | 31,698 | −20.64% |
| pfson_conon_inson | 39,168 | −16.97% | 32,280 | −15.05% | 30,834 | −22.81% |
PolarDB for MySQL
| Scenario | Read-only QPS | vs. baseline | Read/write QPS | vs. baseline | Write-only QPS | vs. baseline |
|---|---|---|---|---|---|---|
| pfsoff | 36,768 | Baseline | 38,876 | Baseline | 39,550 | Baseline |
| pfson_conoff_insoff | 31,981 | −13.02% | 33,342 | −14.24% | 34,182 | −13.57% |
| pfson_conon_insoff | 31,187 | −15.18% | 32,980 | −15.17% | 31,768 | −19.68% |
| pfson_conon_ins_sqlon | 30,976 | −15.75% | 31,980 | −17.77% | 31,101 | −21.36% |
| pfson_conoff_inson | 31,025 | −15.62% | 32,855 | −15.49% | 31,750 | −19.72% |
| pfson_conon_inson | 29,652 | −19.35% | 30,656 | −21.14% | 30,014 | −24.11% |
Test data
Average QPS across all concurrency levels (1 to 1,024), measured after CPU utilization reaches 100%.
ApsaraDB RDS for MySQL
Read-only scenario
Table 1. Test data in the read-only scenario
| Concurrency level | pfsoff | pfson_conoff_insoff | pfson_conon_insoff | pfson_conon_ins_sqlon | pfson_conoff_inson | pfson_conon_inson |
|---|---|---|---|---|---|---|
| 1 | 3,456 (baseline) | 3,209 (−7.16%) | 3,233 (−6.45%) | 3,179 (−8.02%) | 3,249 (−5.98%) | 3,116 (−9.84%) |
| 2 | 9,432 (baseline) | 8,668 (−8.10%) | 8,701 (−7.75%) | 8,588 (−8.95%) | 8,854 (−6.13%) | 8,313 (−11.86%) |
| 4 | 16,485 (baseline) | 14,965 (−9.22%) | 15,123 (−8.26%) | 14,926 (−9.46%) | 14,942 (−9.36%) | 14,053 (−14.75%) |
| 8 | 25,614 (baseline) | 23,024 (−10.11%) | 23,286 (−9.09%) | 22,781 (−11.06%) | 23,017 (−10.14%) | 21,659 (−15.44%) |
| 16 | 39,846 (baseline) | 35,869 (−9.98%) | 35,483 (−10.95%) | 34,162 (−13.14%) | 34,288 (−13.95%) | 33,387 (−16.21%) |
| 32 | 47,176 (baseline) | 41,111 (−12.86%) | 41,773 (−11.45%) | 39,889 (−15.44%) | 39,344 (−16.61%) | 39,168 (−16.97%) |
| 64 | 49,230 (baseline) | 42,186 (−14.31%) | 42,776 (−13.11%) | 41,791 (−15.11%) | 41,747 (−15.20%) | 38,877 (−21.03%) |
| 128 | 48,201 (baseline) | 41,458 (−13.99%) | 41,280 (−14.36%) | 40,797 (−15.36%) | 40,990 (−14.96%) | 36,532 (−24.21%) |
| 256 | 46,982 (baseline) | 40,090 (−14.67%) | 39,869 (−15.14%) | 39,556 (−15.81%) | 40,414 (−13.98%) | 34,602 (−26.35%) |
| 512 | 47,026 (baseline) | 41,279 (−15.22%) | 40,353 (−14.19%) | 39,488 (−16.03%) | 39,337 (−16.35%) | 34,023 (−27.65%) |
| 1,024 | 47,533 (baseline) | 40,104 (−15.63%) | 40,032 (−15.78%) | 39,509 (−16.88%) | 39,462 (−16.98%) | 33,696 (−29.11%) |
Write-only scenario
Table 2. Test data in the write-only scenario
| Concurrency level | pfsoff | pfson_conoff_insoff | pfson_conon_insoff | pfson_conon_ins_sqlon | pfson_conoff_inson | pfson_conon_inson |
|---|---|---|---|---|---|---|
| 1 | 3,614 (baseline) | 3,305 (−8.56%) | 3,293 (−8.88%) | 3,269 (−9.56%) | 3,354 (−7.19%) | 3,222 (−10.86%) |
| 2 | 8,452 (baseline) | 7,644 (−9.56%) | 7,547 (−10.71%) | 7,606 (−10.01%) | 7,603 (−10.04%) | 7,281 (−13.86%) |
| 4 | 13,648 (baseline) | 12,231 (−10.38%) | 12,193 (−10.66%) | 12,106 (−11.30%) | 12,057 (−11.66%) | 11,676 (−14.45%) |
| 8 | 25,648 (baseline) | 22,227 (−13.34%) | 22,824 (−11.01%) | 22,360 (−12.82%) | 22,534 (−12.14%) | 21,431 (−16.44%) |
| 16 | 32,154 (baseline) | 27,119 (−15.66%) | 27,669 (−13.95%) | 27,093 (−15.74%) | 27,347 (−14.95%) | 26,170 (−18.61%) |
| 32 | 39,940 (baseline) | 32,327 (−19.06%) | 31,434 (−21.30%) | 31,865 (−20.22%) | 31,698 (−20.64%) | 30,834 (−22.81%) |
| 64 | 39,123 (baseline) | 30,915 (−20.98%) | 31,220 (−20.20%) | 30,661 (−21.63%) | 30,672 (−21.06%) | 29,514 (−24.56%) |
| 128 | 41,320 (baseline) | 32,692 (−20.88%) | 32,610 (−21.08%) | 32,135 (−22.23%) | 32,184 (−22.11%) | 29,423 (−28.79%) |
| 256 | 42,004 (baseline) | 33,322 (−20.67%) | 33,124 (−21.14%) | 32,838 (−21.82%) | 32,814 (−21.88%) | 30,348 (−27.75%) |
| 512 | 43,156 (baseline) | 34,033 (−21.14%) | 33,597 (−22.15%) | 33,783 (−21.72%) | 33,196 (−23.08%) | 30,157 (−30.12%) |
| 1,024 | 47,562 (baseline) | 37,056 (−22.09%) | 36,232 (−23.82%) | 37,113 (−21.97%) | 36,723 (−22.79%) | 32,256 (−32.18%) |
Read/write scenario
Table 3. Test data in the read/write scenario
| Concurrency level | pfsoff | pfson_conoff_insoff | pfson_conon_insoff | pfson_conon_ins_sqlon | pfson_conoff_inson | pfson_conon_inson |
|---|---|---|---|---|---|---|
| 1 | 2,997 (baseline) | 2,754 (−8.1%) | 2,844 (−5.12%) | 2,758 (−7.98%) | 2,741 (−8.55%) | 2,672 (−10.86%) |
| 2 | 8,135 (baseline) | 7,304 (−10.22%) | 7,610 (−6.45%) | 7,420 (−8.79%) | 7,498 (−7.83%) | 7,226 (−11.17%) |
| 4 | 14,568 (baseline) | 13,031 (−10.55%) | 13,427 (−7.83%) | 13,178 (−9.54%) | 13,164 (−9.64%) | 12,754 (−12.45%) |
| 8 | 24,658 (baseline) | 21,933 (−11.05%) | 22,663 (−8.09%) | 21,854 (−11.37%) | 22,130 (−10.25%) | 21,250 (−13.82%) |
| 16 | 31,562 (baseline) | 28,119 (−10.91%) | 28,443 (−9.88%) | 27,629 (−12.46%) | 27,936 (−11.49%) | 26,957 (−14.61%) |
| 32 | 37,999 (baseline) | 33,170 (−12.71%) | 33,568 (−11.66%) | 33,056 (−13.01%) | 32,698 (−13.95%) | 32,280 (−15.05%) |
| 64 | 39,120 (baseline) | 33,311 (−14.85%) | 33,948 (−13.22%) | 33,068 (−15.47%) | 33,659 (−13.96%) | 31,605 (−19.21%) |
| 128 | 41,687 (baseline) | 35,742 (−14.26%) | 35,334 (−15.24%) | 35,451 (−14.96%) | 35,488 (−14.87%) | 32,899 (−21.08%) |
| 256 | 42,654 (baseline) | 36,243 (−15.03%) | 33,570 (−16.14%) | 36,115 (−15.33%) | 35,949 (−15.72%) | 33,133 (−22.32%) |
| 512 | 43,156 (baseline) | 36,626 (−15.13%) | 36,087 (−16.38%) | 36,302 (−15.88%) | 36,286 (−15.92%) | 32,518 (−24.65%) |
| 1,024 | 43,658 (baseline) | 36,699 (−15.94%) | 36,625 (−16.11%) | 36,585 (−16.20%) | 36,756 (−15.81%) | 32,359 (−25.88%) |
PolarDB for MySQL
Read-only scenario
Table 4. Test data in the read-only scenario
| Concurrency level | pfsoff | pfson_conoff_insoff | pfson_conon_insoff | pfson_conon_ins_sqlon | pfson_conoff_inson | pfson_conon_inson |
|---|---|---|---|---|---|---|
| 1 | 3,389 (baseline) | 3,296 (−2.74%) | 3,233 (−4.62%) | 3,179 (−2.56%) | 3,249 (−4.13%) | 3,016 (−11.04%) |
| 2 | 8,868 (baseline) | 8,632 (−3.08%) | 8,601 (−3.78%) | 8,588 (−3.16%) | 8,654 (−2.84%) | 8,213 (−7.39%) |
| 4 | 16,965 (baseline) | 14,485 (−14.62%) | 15,123 (−10.86%) | 14,926 (−12.02%) | 14,942 (−11.92%) | 13,053 (−23.06%) |
| 8 | 25,024 (baseline) | 22,614 (−9.63%) | 23,286 (−9.01%) | 22,781 (−8.96%) | 23,017 (−8.14%) | 21,659 (−14.45%) |
| 16 | 38,869 (baseline) | 35,846 (−9.88%) | 35,483 (−8.95%) | 34,162 (−12.11%) | 34,288 (−11.79%) | 32,387 (−16.68%) |
| 32 | 36,768 (baseline) | 31,981 (−13.02%) | 31,187 (−15.18%) | 30,976 (−15.75%) | 31,025 (−15.62%) | 29,652 (−19.35%) |
| 64 | 49,186 (baseline) | 42,230 (−14.14%) | 42,776 (−13.07%) | 41,791 (−15.03%) | 41,747 (−15.12%) | 39,877 (−18.93%) |
| 128 | 42,958 (baseline) | 39,201 (−11.88%) | 40,280 (−9.08%) | 39,797 (−10.43%) | 39,990 (−10.11%) | 37,532 (−12.63%) |
| 256 | 45,090 (baseline) | 41,026 (−9.01%) | 39,869 (−11.58%) | 39,556 (−12.27%) | 40,414 (−10.37%) | 35,602 (−21.04%) |
| 512 | 46,279 (baseline) | 41,004 (−11.14%) | 40,353 (−12.81%) | 39,488 (−14.67%) | 39,337 (−15.02%) | 34,023 (−26.48%) |
| 1,024 | 47,231 (baseline) | 41,533 (−12.06%) | 40,032 (−15.24%) | 39,509 (−16.35%) | 39,462 (−16.45%) | 33,696 (−28.66%) |
Write-only scenario
Table 5. Test data in the write-only scenario
| Concurrency level | pfsoff | pfson_conoff_insoff | pfson_conon_insoff | pfson_conon_ins_sqlon | pfson_conoff_inson | pfson_conon_inson |
|---|---|---|---|---|---|---|
| 1 | 3,505 (baseline) | 3,364 (−4.07%) | 3,354 (−5.41%) | 3,269 (−6.73%) | 3,293 (−6.05%) | 3,122 (−10.93%) |
| 2 | 7,844 (baseline) | 7,452 (−4.56%) | 7,603 (−4.22%) | 7,606 (−5.09%) | 7,547 (−3.44%) | 7,281 (−7.18%) |
| 4 | 13,231 (baseline) | 12,648 (−4.38%) | 12,057 (−8.87%) | 12,106 (−8.30%) | 12,193 (−7.85%) | 10,676 (−19.31%) |
| 8 | 25,227 (baseline) | 22,648 (−10.22%) | 22,534 (−10.68%) | 22,360 (−11.36%) | 22,824 (−9.53%) | 20,431 (−19.01%) |
| 16 | 27,119 (baseline) | 26,154 (−13.66%) | 26,347 (−11.95%) | 25,093 (−14.74%) | 26,669 (−12.95%) | 22,170 (−18.25%) |
| 32 | 39,550 (baseline) | 34,182 (−13.57%) | 31,768 (−19.68%) | 31,101 (−21.36%) | 31,750 (−19.72%) | 30,014 (−24.11%) |
| 64 | 37,915 (baseline) | 31,223 (−17.65%) | 30,672 (−19.13%) | 30,661 (−19.12%) | 31,220 (−17.66%) | 29,514 (−22.16%) |
| 128 | 34,692 (baseline) | 31,320 (−16.88%) | 31,884 (−12.08%) | 32,135 (−10.76%) | 31,910 (−14.11%) | 29,423 (−15.22%) |
| 256 | 33,390 (baseline) | 31,704 (−7.67%) | 31,814 (−7.14%) | 32,018 (−5.41%) | 31,124 (−9.88%) | 29,348 (−12.11%) |
| 512 | 34,033 (baseline) | 31,156 (−8.45%) | 31,196 (−8.34%) | 31,783 (−6.61%) | 31,597 (−7.16%) | 30,157 (−11.39%) |
| 1,024 | 36,855 (baseline) | 33,862 (−8.12%) | 33,723 (−8.52%) | 33,613 (−8.81%) | 33,232 (−9.83%) | 32,956 (−10.58%) |
Read/write scenario
Table 6. Test data in the read/write scenario
| Concurrency level | pfsoff | pfson_conoff_insoff | pfson_conon_insoff | pfson_conon_ins_sqlon | pfson_conoff_inson | pfson_conon_inson |
|---|---|---|---|---|---|---|
| 1 | 3,305 (baseline) | 2,754 (−16.67%) | 2,843 (−13.98%) | 2,758 (−16.55%) | 2,741 (−17.07%) | 2,614 (−20.91%) |
| 2 | 8,668 (baseline) | 7,304 (−15.74%) | 7,610 (−12.21%) | 7,420 (−14.42%) | 7,498 (−13.52%) | 7,182 (−17.14%) |
| 4 | 14,965 (baseline) | 13,031 (−12.92%) | 13,427 (−10.28%) | 13,178 (−11.94%) | 13,164 (−12.03%) | 12,754 (−14.77%) |
| 8 | 23,068 (baseline) | 21,933 (−9.94%) | 21,663 (−6.09%) | 21,354 (−8.68%) | 21,130 (−8.41%) | 20,250 (−12.22%) |
| 16 | 35,869 (baseline) | 28,119 (−14.96%) | 28,443 (−17.71%) | 28,629 (−16.74%) | 27,936 (−16.95%) | 26,957 (−24.85%) |
| 32 | 38,876 (baseline) | 33,342 (−14.24%) | 32,980 (−15.17%) | 31,980 (−17.74%) | 32,855 (−15.49%) | 30,656 (−21.14%) |
| 64 | 42,186 (baseline) | 33,311 (−20.86%) | 33,948 (−19.53%) | 33,068 (−21.61%) | 33,659 (−20.21%) | 31,305 (−25.79%) |
| 128 | 40,359 (baseline) | 35,742 (−11.44%) | 35,334 (−12.45%) | 35,451 (−12.16%) | 35,488 (−12.07%) | 32,899 (−18.48%) |
| 256 | 40,090 (baseline) | 36,243 (−9.67%) | 33,570 (−16.26%) | 36,115 (−9.92%) | 35,949 (−10.33%) | 33,125 (−17.37%) |
| 512 | 41,220 (baseline) | 36,626 (−11.15%) | 36,087 (−12.45%) | 36,302 (−11.93%) | 36,286 (−11.98%) | 32,451 (−20.27%) |
| 1,024 | 40,055 (baseline) | 36,699 (−8.38%) | 36,625 (−8.56%) | 36,585 (−8.66%) | 36,756 (−8.24%) | 32,860 (−17.96%) |