OCP Performance Analyzer MCP - Interactive Demo

👋 Welcome to OpenShift ETCD Performance Analyzer

I can help you analyze OpenShift cluster performance, check system health, and provide AI-powered deep insights.

Click the Quick Actions buttons on the left to start the demo

🏗️ System Architecture Design

┌─────────────────────────────────────────────────────────────┐
│                       Client Layer                          │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐  ┌──────────┐   │
│  │  Web UI  │  │ CLI Tools│  │ REST API │  │MCP Client│   │
│  └────┬─────┘  └────┬─────┘  └────┬─────┘  └────┬─────┘   │
└───────┼─────────────┼─────────────┼─────────────┼──────────┘
        └─────────────┼─────────────┼─────────────┘
                      │             │
┌─────────────────────┼─────────────┼──────────────────────────┐
│            AI Agent Layer (FastAPI + LangGraph)              │
│  ┌────────────────────────────────────────────────────────┐ │
│  │  🤖 Chat Agent  (Conversational Interaction)           │ │
│  │  📊 Report Agent (Auto Report Generation)              │ │
│  │  💾 Storage Agent (Data Persistence)                   │ │
│  │  • Streaming Response (SSE)  • Tool Orchestration      │ │
│  │  • Context Memory                                      │ │
│  └────────────────────────────────────────────────────────┘ │
└─────────────────────┬─────────────┬──────────────────────────┘
                      │ MCP Protocol Communication │
┌─────────────────────┼─────────────┼──────────────────────────┐
│              MCP Server Layer (FastMCP Framework)            │
│  ┌─────────────────────────────────────────────────────────┐│
│  │  ETCD Analyzer Server (Port 8001)                       ││
│  │  • 15 Analysis Tools                                    ││
│  │  • Cluster Status  • WAL Fsync  • Backend Commit        ││
│  │  • Disk I/O   • Network I/O    • Deep Analysis          ││
│  └─────────────────────────────────────────────────────────┘│
└─────────────────────┬─────────────┬──────────────────────────┘
                      │             │
┌─────────────────────┼─────────────┼──────────────────────────┐
│                Tools Layer (Tools & Collectors)              │
│  ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐       │
│  │Prometheus│ │  PromQL  │ │  Metric  │ │   Data   │       │
│  │Integration│ │  Query   │ │Collector │ │Transform │       │
│  │          │ │ Execution│ │  (51+)   │ │  (ELT)   │       │
│  └──────────┘ └──────────┘ └──────────┘ └──────────┘       │
└─────────────────────┬─────────────┬──────────────────────────┘
                      │             │
┌─────────────────────┼─────────────┼──────────────────────────┐
│              Analysis Layer (Analysis & Storage)             │
│  ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐       │
│  │Bottleneck│ │Performance│ │  DuckDB  │ │   HTML   │       │
│  │Detection │ │ Analysis │ │  Storage │ │  Report  │       │
│  │  Engine  │ │Algorithm │ │(TimeSeries)│ │Generation│       │
│  └──────────┘ └──────────┘ └──────────┘ └──────────┘       │
└──────────────────────────────────────────────────────────────┘
                      │
┌─────────────────────┼──────────────────────────────────────────┐
│          OpenShift/Kubernetes Cluster Infrastructure         │
│  • ETCD Cluster (3-5 nodes)  • Prometheus/Thanos Monitoring │
│  • Master Nodes              • Kubernetes API Server        │
│  • OVN-Kubernetes Network    • Container Runtime            │
└──────────────────────────────────────────────────────────────┘

🔧 Key Component Description

1. MCP Server (FastMCP) - Model Context Protocol-based server exposing 15+ analysis tools for AI agent invocation

2. AI Agent (LangGraph) - Intelligent agent using OpenAI/compatible LLM for conversational analysis and automated report generation

3. Metric Collector - Queries Prometheus via PromQL to collect 51+ ETCD performance metrics

4. ELT Pipeline - Transforms raw JSON data into structured HTML tables for easy display and analysis

5. DuckDB Storage - Time-series database storing historical performance data for trend analysis

🎯 Core Features Demo

1️⃣ Real-time Performance Monitoring

                        Feature: Real-time ETCD cluster performance metrics collection via Prometheus

                        Metric Coverage: 51 core ETCD metrics covering disk, network, CPU, memory, etc.

                        Update Frequency: 2-minute sampling window with real-time streaming response

2️⃣ AI Conversational Analysis

# User Query Example
User: "Analyze ETCD WAL fsync performance over the past 1 hour"

# AI Execution Flow
1. Call Tool: get_etcd_disk_wal_fsync(duration="1h")
2. Collect Metrics: P50/P90/P99 latency distribution
3. Analyze Data: Compare against thresholds (target: P99 < 10ms)
4. Generate Recommendations: Provide optimization suggestions if thresholds exceeded

# Response
📊 Data Result:
  - WAL Fsync P99: 8.2ms ✓ (Excellent)
  - WAL Fsync P90: 5.1ms
  - WAL Fsync P50: 2.3ms

🤖 AI Analysis:
  WAL fsync performance is excellent, P99 latency of 8.2ms is well below 10ms threshold.
  Disk write performance is stable, no optimization needed.
                    

3️⃣ Automatic Bottleneck Detection

                        Detection Dimensions:

                        • CPU Usage (Threshold: Warning 70%, Critical 85%)

                        • Memory Usage (Threshold: Warning 70%, Critical 85%)

                        • WAL Fsync Latency (Target: P99 < 10ms)

                        • Backend Commit Latency (Target: P99 < 25ms)

                        • Disk I/O Wait (Target: < 10%)

                        • Network Latency (Target: Peer < 50ms)

4️⃣ Complete Performance Report

# Report Structure
┌─────────────────────────────────────┐
│  📋 ETCD Performance Analysis Report│
│  Time Range: 2026-04-12 14:00-15:00│
├─────────────────────────────────────┤
│  1. Executive Summary               │
│     • Cluster Status: Healthy ✓     │
│     • Issues Found: 1 warning       │
│     • Priority: Medium              │
├─────────────────────────────────────┤
│  2. Key Metrics                     │
│     ✓ WAL Fsync P99: 8.2ms          │
│     ⚠ Backend Commit P99: 32.5ms    │
│     ✓ CPU Usage: 45%                │
│     ✓ Memory Usage: 58%             │
│     ✗ Disk I/O Wait: 15%            │
├─────────────────────────────────────┤
│  3. Bottleneck Analysis             │
│     🔴 Disk I/O Bottleneck          │
│        - I/O Wait 15% too high      │
│        - Affects Backend Commit     │
├─────────────────────────────────────┤
│  4. Optimization Recommendations    │
│     1. Use faster SSD (NVMe)        │
│     2. Tune I/O scheduler to deadline│
│     3. Check disk contention        │
└─────────────────────────────────────┘
                    

🔍 Tool List (15 Tools)

Tool Name	Function Description	Return Data
get_server_health	Server health check	Status, timestamp
get_etcd_cluster_status	ETCD cluster status	Member list, health status
get_ocp_cluster_info	OpenShift cluster info	Version, node count, resources
get_etcd_general_info	ETCD general metrics	CPU, memory, DB size
get_etcd_disk_wal_fsync	WAL Fsync performance	P50/P90/P99 latency
get_etcd_disk_backend_commit	Backend Commit performance	P50/P90/P99 latency
get_node_disk_io	Node disk I/O	IOPS, throughput, latency
get_etcd_network_io	Network I/O metrics	Bandwidth, packet rate
get_etcd_performance_deep_drive	Deep performance analysis	Multi-dimensional analysis
get_etcd_bottleneck_analysis	Bottleneck detection	Bottleneck identification & suggestions
generate_etcd_performance_report	Generate performance report	Complete HTML report

🚀 OCP ETCD Analyzer

📊 Quick Actions

Performance Analysis Demo