Agent Skills: Optimize Throttles

Optimize Throttles

Analyze throttle behavior and autonomously configure optimal throttle settings to balance workload priorities, prevent resource starvation, and meet service level agreements.

🎯 New Autonomous Capabilities

This skill can now execute changes, not just recommend them!

With tdwm-mcp v1.5.0, this skill has evolved from advisory to autonomous:

• ✅ CREATE new throttles based on analysis
• ✅ MODIFY existing throttle limits dynamically
• ✅ ENABLE/DISABLE throttles without manual SQL
• ✅ USE TEMPLATES for common throttle patterns
• ✅ VALIDATE parameters before execution
• ✅ VERIFY changes were applied correctly

Instructions

When to Use This Skill

• Users report excessive query delays or throttling
• Need to balance competing workload priorities
• Workload SLAs are not being met
• After adding new workloads or applications
• Following system capacity changes
• System overload requiring immediate throttle creation
• Seasonal workload changes need different limits

Available MCP Tools

Monitoring & Analysis:

• show_trottle_statistics - View throttle behavior by level (ALL/QUERY/SESSION/WORKLOAD)
• show_tasm_statistics - Review TASM rule effectiveness
• show_tdwm_summary - See workload distribution
• list_active_WD - List active workload definitions
• show_query_log - Analyze query timing and delays
• list_rulesets - List all available rulesets

Configuration Management (NEW ✨):

• create_system_throttle - Create new throttle with classifications
• modify_throttle_limit - Dynamically adjust throttle limits
• delete_throttle - Remove throttle definition
• enable_throttle - Activate throttle rule
• disable_throttle - Deactivate throttle rule
• add_classification_to_rule - Add classification criteria
• add_subcriteria_to_target - Add sub-criteria (FTSCAN, MINSTEPTIME, etc.)
• activate_ruleset - Apply all pending changes

Available MCP Resources (NEW ✨)

Templates:

• tdwm://templates/throttle - List all throttle templates
• tdwm://template/throttle/application-basic - Limit queries by application
• tdwm://template/throttle/table-fullscan - Limit full table scans
• tdwm://template/throttle/user-concurrency - Per-user concurrency limits
• tdwm://template/throttle/time-based-etl - Time-based ETL throttling

Reference Data:

• tdwm://reference/classification-types - All 31 classification types
• tdwm://reference/operators - Classification operators (I, O, IO)
• tdwm://reference/subcriteria-types - Sub-criteria types
• tdwm://reference/throttle-types - Throttle types (DM, M)

Discovery:

• tdwm://rulesets - List all rulesets
• tdwm://system/active-ruleset - Get currently active ruleset
• tdwm://ruleset/{name}/throttles - List throttles in ruleset
• tdwm://ruleset/{name}/throttle/{throttle_name} - Throttle details
• tdwm://ruleset/{name}/pending-changes - Check pending changes

Workflows:

• tdwm://workflow/create-throttle - Step-by-step throttle creation guide
• tdwm://workflow/emergency-throttle - Emergency response workflow

Step-by-Step Workflow

Phase 1: Discovery & Analysis (Read-Only)

1. Assess Current Throttle Behavior

   • Use show_trottle_statistics for all levels (query/session/workload)
   • Identify workloads experiencing excessive throttling
   • Note throttle delay times and frequency

2. Understand Workload Requirements

   • Review business SLAs for each workload
   • Identify critical vs non-critical workloads
   • Determine acceptable delay tolerances

3. Analyze Workload Interactions

   • Use show_tdwm_summary to see resource competition
   • Identify which workloads are blocking others
   • Check if high-priority workloads are being throttled

4. Review Current Configuration

   • Get active ruleset: tdwm://system/active-ruleset
   • List current throttles: tdwm://ruleset/{name}/throttles
   • For each throttle: tdwm://ruleset/{name}/throttle/{throttle_name}
   • Identify mismatches between priority and throttle limits

5. Calculate Optimal Settings

   • Based on resource availability and workload requirements
   • Consider peak vs average usage patterns
   • Balance protection (prevent monopolization) vs throughput
   • Factor in concurrency needs for each workload

Phase 2: Execute Changes (Autonomous - NEW!)

6. Select Approach: Template or Custom

   • Template-Based (Recommended):

     • Browse templates: tdwm://templates/throttle
     • Select template matching use case
     • Read template: tdwm://template/throttle/{id}
     • Customize parameters for your workload

   • Custom Throttle:

     • Validate classification types: tdwm://reference/classification-types
     • Validate operators: tdwm://reference/operators
     • Design custom classification criteria

7. Create or Modify Throttle

   • For NEW throttles:

     Use: create_system_throttle(
       ruleset_name, throttle_name, throttle_type,
       limit, classification_criteria
     )
     

   • For EXISTING throttles:

     Use: modify_throttle_limit(
       ruleset_name, throttle_name, new_limit
     )
     

8. Add Additional Criteria (if needed)

   • Add classifications: add_classification_to_rule
   • Add sub-criteria: add_subcriteria_to_target (FTSCAN, MINSTEPTIME, etc.)

9. Enable and Activate

   • Enable throttle: enable_throttle(ruleset_name, throttle_name)
   • Check pending changes: tdwm://ruleset/{name}/pending-changes
   • Activate ruleset: activate_ruleset(ruleset_name)

Phase 3: Verification (Autonomous - NEW!)

10. Verify Configuration

    • Read back throttle: tdwm://ruleset/{name}/throttle/{throttle_name}
    • Confirm limit matches expected value
    • Confirm enabled state is correct

11. Monitor Impact

    • Wait 5-10 minutes for statistics to update
    • Check throttle statistics: show_trottle_statistics
    • Verify delayed query count is acceptable
    • Monitor workload summary: show_tdwm_summary

12. Iterate if Needed

    • If throttling is too aggressive: increase limit
    • If not effective enough: decrease limit
    • Use modify_throttle_limit for adjustments

Examples

Example 1: Create ETL Throttle to Protect Interactive Queries

Scenario: Batch ETL jobs overwhelming system, slowing interactive reports

Discovery:

1. show_tdwm_summary → See ETL using 80% of resources
2. show_trottle_statistics → No ETL throttle exists
3. tdwm://system/active-ruleset → "Tactical"

Analysis:

• ETL should run max 10 concurrent queries
• Interactive reports need protection
• ETL workload identified by APPL='ETL_APP'

Execution (Autonomous):

1. Select template:
   tdwm://template/throttle/application-basic

2. Create throttle:
   create_system_throttle(
     ruleset_name="Tactical",
     throttle_name="ETL_LIMIT",
     throttle_type="DM",  # Delay Management
     limit=10,
     classification_criteria=[{
       "description": "ETL Application throttle",
       "type": "APPL",
       "value": "ETL_APP",
       "operator": "I"  # Inclusion
     }]
   )

3. Enable:
   enable_throttle(ruleset_name="Tactical", throttle_name="ETL_LIMIT")

4. Activate:
   activate_ruleset(ruleset_name="Tactical")

5. Verify:
   tdwm://ruleset/Tactical/throttle/ETL_LIMIT
   → Confirm limit=10, enabled=true

6. Monitor:
   show_trottle_statistics(type="ALL")
   → Check ETL delayed count after 10 minutes

Result: ETL limited to 10 concurrent, interactive reports responsive

Example 2: Adjust Existing Throttle Limit

Scenario: Analytics throttle too restrictive (limit=3), queries backing up

Discovery:

1. show_trottle_statistics → Analytics delayed count growing
2. tdwm://ruleset/Tactical/throttle/ANALYTICS_LIMIT → Current limit=3
3. show_query_log → Analytics queries delayed avg 15 minutes

Analysis:

• Limit of 3 is too low for current analytics load
• Recommend increasing to 8 based on workload analysis
• System has capacity for increase

Execution (Autonomous):

1. Modify limit:
   modify_throttle_limit(
     ruleset_name="Tactical",
     throttle_name="ANALYTICS_LIMIT",
     new_limit=8
   )

2. Activate:
   activate_ruleset(ruleset_name="Tactical")

3. Verify:
   tdwm://ruleset/Tactical/throttle/ANALYTICS_LIMIT
   → Confirm limit=8

4. Monitor:
   show_trottle_statistics(type="ALL")
   → Delayed count should decrease

Result: Analytics throughput increased, delay reduced to <5 min

Example 3: Emergency Throttle During System Overload

Scenario: Unexpected query spike causing CPU >95%, need immediate restriction

Analysis: SKIP detailed analysis, immediate action needed

Execution (Fast):

1. Use emergency workflow:
   tdwm://workflow/emergency-throttle

2. Create very restrictive throttle:
   create_system_throttle(
     ruleset_name="Tactical",
     throttle_name="EMERGENCY_LIMIT",
     throttle_type="DM",
     limit=3,  # Very restrictive
     classification_criteria=[{
       "description": "Emergency system protection",
       "type": "APPL",
       "value": "*",  # ALL applications
       "operator": "I"
     }]
   )

3. Enable and activate immediately:
   enable_throttle(ruleset_name="Tactical", throttle_name="EMERGENCY_LIMIT")
   activate_ruleset(ruleset_name="Tactical")

4. Monitor recovery:
   Wait 60 seconds, check show_physical_resources
   → CPU should drop below 90%

5. Gradual relaxation:
   modify_throttle_limit(..., new_limit=5)
   activate_ruleset(...)
   Wait, monitor

   modify_throttle_limit(..., new_limit=10)
   activate_ruleset(...)
   Wait, monitor

6. Cleanup when stable:
   delete_throttle(ruleset_name="Tactical", throttle_name="EMERGENCY_LIMIT")
   activate_ruleset(ruleset_name="Tactical")

Result: System stabilized in <5 minutes, gradual return to normal

Example 4: Template-Driven Full Table Scan Throttle

Scenario: Full table scans causing AMP skew and resource contention

Discovery:

1. show_amp_load → High skew factor
2. show_query_log → Many queries doing full table scans
3. Need to limit FTSCAN queries

Execution (Template-Based):

1. Get template:
   tdwm://template/throttle/table-fullscan
   → Shows structure with FTSCAN sub-criteria

2. Create throttle using template pattern:
   create_system_throttle(
     ruleset_name="Tactical",
     throttle_name="FTSCAN_LIMIT",
     throttle_type="DM",
     limit=5,
     classification_criteria=[{
       "description": "Target tables",
       "type": "TABLE",
       "value": "LARGE_TABLE_*",  # Pattern matching
       "operator": "I"
     }]
   )

3. Add FTSCAN sub-criteria:
   add_subcriteria_to_target(
     ruleset_name="Tactical",
     throttle_name="FTSCAN_LIMIT",
     target_type="TABLE",
     subcriteria={
       "type": "FTSCAN",
       "value": "Y"  # Full scan = Yes
     }
   )

4. Enable and activate:
   enable_throttle(ruleset_name="Tactical", throttle_name="FTSCAN_LIMIT")
   activate_ruleset(ruleset_name="Tactical")

5. Verify:
   tdwm://ruleset/Tactical/throttle/FTSCAN_LIMIT
   → Confirm sub-criteria present

Result: Full table scans limited, AMP skew reduced

Best Practices

General Principles

• Throttles should reflect business priorities, not just technical factors
• Set throttles to prevent monopolization while allowing throughput
• High-priority workloads need guaranteed minimum resources
• Low-priority workloads should have limits to prevent starvation of others
• Always leave some headroom for unexpected workload spikes
• Document throttle rationale for future reference

Discovery Before Execution (NEW ✨)

• ALWAYS use tdwm://ruleset/{name}/throttles to see existing throttles before creating
• Check tdwm://ruleset/{name}/pending-changes before activating
• Verify with tdwm://ruleset/{name}/throttle/{name} after changes

Template Usage (NEW ✨)

• Use templates for common patterns (reduces errors by 50%)
• Customize template parameters for your specific needs
• Follow template best practices and examples

Execution Safety (NEW ✨)

• Start with conservative limits, increase gradually
• Test throttle changes in non-production first if possible
• Monitor impact for 15-30 minutes before declaring success
• Keep emergency rollback plan (delete_throttle + activate_ruleset)

Validation (NEW ✨)

• Validate classification types against tdwm://reference/classification-types
• Validate operators against tdwm://reference/operators
• Check throttle exists after creation (resource read)

Monitoring

• Query-level throttles control concurrency, workload-level control total impact
• Consider different throttle settings for peak vs off-peak hours
• Monitor throttle effectiveness after changes - iterate if needed
• Coordinate throttle changes with workload owners

Related Skills

• Use configure-throttles skill for new throttle creation from scratch
• Use emergency-response skill for crisis situations
• Use discover-configuration skill to inventory existing throttles
• Use tune-workloads skill to adjust classification criteria