Monitoring CPU Usage

graph TD
    A[High CPU 70-90%<br/>Queries slowing down] --> B{What's the cause?}

    B --> C[Inefficient Queries?]
    B --> D[Too Many Connections?]
    B --> E[Missing Indexes?]

    C --> C1["Run PROFILE"]
    C1 --> C2{Check execution plan}
    C2 --> C3["NodeByLabelScan = millions of hits<br/>AllNodesScan = scanning everything<br/>Expand with high DB hits"]
    C3 --> C4["Optimize:<br/>Filter early, use selective patterns"]

    D --> D1[Check active connections<br/>vs available cores]
    D1 --> D2{More than 15 per core?}
    D2 -->|Yes| D3["Reduce connections<br/>200 connections on 8 cores = problem<br/>Review pool configuration"]
    D2 -->|No| D4[Connections OK]

    E --> E1["Look in PROFILE results:<br/>Filter after retrieval?<br/>No NodeIndexSeek?"]
    E1 --> E2["CREATE INDEX person_name<br/>FOR (p:Person) ON (p.name)"]
    E2 --> E3["Restructure query:<br/>MATCH (f:Person {name: $name})-[:KNOWS]-(p)<br/>Use parameters for plan caching"]

    C4 --> F{Did CPU drop?}
    D3 --> F
    D4 --> F
    E3 --> F

    F -->|Yes| G["Success!<br/>Monitor and maintain"]
    F -->|No| H["All optimizations done?<br/>Time to scale instance"]

graph TD
    A["CPU Spiking Regularly<br/>All queries degraded at once"] --> B{Check GC Metrics First}

    B -->|"GC time >5%<br/>⚠️ Critical"| C["Memory Pressure Problem<br/>JVM spending too much time<br/>cleaning up memory"]
    B -->|"GC time 1-5%<br/>⚠ Warning"| D["Borderline - investigate<br/>before it gets worse"]
    B -->|"GC time <1%<br/>✓ Normal"| E["GC is fine<br/>Look for other causes"]

    C --> C1["Check heap utilization<br/>in metrics dashboard"]
    C1 --> C2{Heap consistently<br/>above 90%?}
    C2 -->|Yes| C3["Solution: Increase instance size<br/>You need more memory,<br/>not query optimization"]
    C2 -->|No| D

    D --> D1["Review query logs<br/>Look at timestamps"]
    D1 --> D2["Do spikes correlate with:<br/>- Specific queries?<br/>- Scheduled jobs?<br/>- Peak traffic times?"]
    D2 --> D3["Optimize those queries<br/>or reschedule operations"]

    E --> E1["Check for batch operations<br/>ETL jobs, reports, backups"]
    E1 --> E2["Solution: Reschedule to<br/>off-peak hours"]

    E --> E3["Identify long-running queries<br/>holding memory"]
    E3 --> E4["Solution: Optimize or<br/>break into smaller chunks"]

    E --> E5["Check for lock contention<br/>Concurrent writes blocking reads"]
    E5 --> E6["Solution: Separate workloads<br/>or adjust timing"]

graph TD
    A["🚨 CPU at 100%<br/>Queries failing, timeouts"] --> B["IMMEDIATE ACTION<br/>Check Query Monitor"]

    B --> C{Spot any<br/>long-running queries?}

    C -->|"Yes<br/>Found the culprit"| D["Terminate runaway queries<br/>Use query monitoring to<br/>kill specific query IDs"]

    C -->|"No<br/>Everything looks normal"| E["Check recent deployments<br/>New code in last hour?<br/>Traffic spike?"]

    D --> F{Did CPU drop<br/>below 90%?}
    E --> F

    F -->|"Yes - Crisis averted"| G["Quick Fixes<br/>Buy time while investigating"]
    F -->|"No - Still maxed out"| H["🚨 EMERGENCY SCALE<br/>Increase instance size NOW<br/>Investigate after service restored"]

    G --> G1["Add query timeouts<br/>SET query.timeout=30s<br/>Prevent future runaway queries"]
    G --> G2["Restrict expensive patterns<br/>Temporarily disable heavy<br/>analytics or reporting"]

    G1 --> I["Root Cause Analysis<br/>Why did this happen?"]
    G2 --> I
    H --> I

    I --> J{Was this a<br/>one-time event or<br/>growing trend?}

    J -->|"One-time spike<br/>e.g. bad query"| K["Apply Pattern 1 fixes:<br/>- PROFILE the bad query<br/>- Add missing indexes<br/>- Optimize query structure"]

    J -->|"Sustained growth<br/>Traffic increasing"| L["Scale instance permanently<br/>Your workload has outgrown<br/>current capacity"]

    K --> M["Preventive Measures<br/>Stop this happening again"]
    L --> M

    M --> M1["Set up connection pooling<br/>Limit connections per app"]
    M --> M2["Implement rate limiting<br/>Protect against traffic surges"]
    M --> M3["Enable query monitoring alerts<br/>Catch problems early"]