[QUESTION]

[RobCarlton]

Which component is your question about?

Full Dashboard

Performance Monitor Version

2.10

Is your question about how it works, or the results?

How the tool works

What's your question?

Should 1 minute intervals be sufficient enough for the SQL Agent job PerformanceMonitor - Collection or will that timeframe vary on the server/databases?

Additional Context

No response

[erikdarlingdata]

1 minute is the right value, and it's deliberate — that interval is the master clock for the dispatcher, not the collection cadence itself.

Here's how it works:

• The PerformanceMonitor - Collection agent job runs collect.scheduled_master_collector every 1 minute.
• That proc reads config.collection_schedule and only runs collectors whose next_run_time <= SYSDATETIME().
• After each collector runs, its next_run_time is bumped by its own frequency_minutes.

So the 1-minute job tick is the resolution — the floor for how often things can be re-evaluated. The actual per-collector cadence is the frequency_minutes column in config.collection_schedule:

Cadence	Collectors
1 min	wait_stats, memory_stats, blocked_process_xml, deadlock_xml, query_snapshots, file_io_stats, memory_grant_stats, cpu_scheduler/utilization, latch/spinlock_stats, tempdb_stats, session_stats, waiting_tasks, running_jobs, configuration_issues_analyzer, memory_pressure_events
2 min	query_stats, query_store, procedure_stats, trace_analysis
5 min	system_health, blocking/deadlock parsers + analyzer, memory_clerks, perfmon_stats, default_trace, plan_cache_stats
60 min	database_size_stats
1440 min (daily)	trace_management, server/database_configuration, server_properties

Implications for tuning:

• Don't lengthen the job interval past 1 minute. That would silently delay every 1-minute collector (most of the high-value ones). If you set the job to 5 minutes, your "1-minute" wait_stats becomes "5-minute" wait_stats.
• If a server is too busy, tune at the collector level, not the job level: bump frequency_minutes for noisy collectors, or set enabled = 0 for ones you don't need. For example, if query_snapshots is too chatty:

  UPDATE config.collection_schedule
  SET frequency_minutes = 5
  WHERE collector_name = N'query_snapshots_collector';

• The job is also self-throttling via max_duration_minutes. If a collector is still running when its next tick is due, the dispatcher skips it rather than piling on.

So to your question directly: yes, 1 minute is sufficient and recommended for every server — the per-server variation lives in config.collection_schedule, not the agent job schedule.