Application Performance Monitoring (APM)

---

1 - Qu'est-ce que l'APM ?

1.1 Définition

APM = Surveillance des performances applicatives en temps réel.

1.2 Métriques clés

Métrique	Description	Cible
Apdex	Score de satisfaction	> 0.9
Response Time	Temps de réponse	< 200ms
Throughput	Requêtes/sec	Variable
Error Rate	% d'erreurs	< 1%
CPU/Memory	Ressources	< 80%

---

2 - Solutions APM

2.1 Comparatif

Solution	Type	Prix	Points forts
Datadog	SaaS	$$$	All-in-one
New Relic	SaaS	$$$	APM natif
Dynatrace	SaaS	$$$$	AI-powered
Elastic APM	OSS/SaaS	$-$$	ELK intégration
Grafana	OSS	$	LGTM stack
SigNoz	OSS	Free	OpenTelemetry

2.2 Architecture open source

---

3 - Elastic APM

3.1 Installation Agent

// Node.js - apm.js
const apm = require('elastic-apm-node').start({
  serviceName: 'my-service',
  serverUrl: 'http://apm-server:8200',
  environment: 'production',
  captureBody: 'all',
  transactionSampleRate: 1.0,
});

module.exports = apm;

# Python
import elasticapm

app = Flask(__name__)
apm = ElasticAPM(app, 
    service_name='my-service',
    server_url='http://apm-server:8200'
)

3.2 Configuration APM Server

# docker-compose.yml
apm-server:
  image: docker.elastic.co/apm/apm-server:8.11.0
  ports:
    - "8200:8200"
  environment:
    - output.elasticsearch.hosts=["elasticsearch:9200"]
    - apm-server.kibana.enabled=true
    - apm-server.kibana.host=kibana:5601

3.3 Custom Transactions

const apm = require('./apm');

// Transaction manuelle
const transaction = apm.startTransaction('process-order', 'custom');

// Span
const span = apm.startSpan('database-query');
await db.query('SELECT * FROM orders');
span.end();

// Labels
apm.setLabel('order_id', orderId);
apm.setLabel('customer_tier', 'premium');

// User context
apm.setUserContext({
  id: user.id,
  username: user.name,
  email: user.email,
});

transaction.end();

---

4 - SigNoz (Open Source)

4.1 Installation

# Docker
git clone https://github.com/SigNoz/signoz.git
cd signoz/deploy
./install.sh

# Kubernetes
helm repo add signoz https://charts.signoz.io
helm install signoz signoz/signoz \
    --namespace monitoring \
    --create-namespace

4.2 Configuration Application

// OpenTelemetry standard - compatible SigNoz
const { NodeSDK } = require('@opentelemetry/sdk-node');
const { OTLPTraceExporter } = require('@opentelemetry/exporter-trace-otlp-http');

const sdk = new NodeSDK({
  traceExporter: new OTLPTraceExporter({
    url: 'http://signoz-otel-collector:4318/v1/traces',
  }),
  // ...
});

---

5 - Profiling

5.1 Continuous Profiling

5.2 Pyroscope

# docker-compose.yml
pyroscope:
  image: pyroscope/pyroscope:latest
  ports:
    - "4040:4040"
  command: ["server"]

// Node.js agent
const Pyroscope = require('@pyroscope/nodejs');

Pyroscope.init({
  serverAddress: 'http://pyroscope:4040',
  appName: 'my-service',
});

Pyroscope.start();

5.3 Flame Graphs

                                      [main]
                    ┌─────────────────────┴──────────────────────┐
              [processRequest]                           [handleResponse]
        ┌───────────┴───────────┐                              │
   [validateInput]        [queryDatabase]                 [serialize]
        │                       │
   [checkAuth]            [executeSQL]

---

6 - Real User Monitoring (RUM)

6.1 Concept

6.2 Implementation

<!-- Elastic RUM -->
<script src="https://unpkg.com/@elastic/apm-rum"></script>
<script>
  elasticApm.init({
    serviceName: 'my-frontend',
    serverUrl: 'https://apm.example.com',
    environment: 'production',
  });
</script>

// React
import { init as initApm } from '@elastic/apm-rum';

const apm = initApm({
  serviceName: 'my-react-app',
  serverUrl: 'https://apm.example.com',
});

// Transaction manuelle
const transaction = apm.startTransaction('user-checkout', 'user-interaction');
// ... checkout logic
transaction.end();

6.3 Métriques RUM

Métrique	Description
LCP	Largest Contentful Paint
FID	First Input Delay
CLS	Cumulative Layout Shift
TTFB	Time to First Byte
FCP	First Contentful Paint

---

7 - Dashboards APM

7.1 Service Overview

panels:
  - title: "Request Rate"
    type: timeseries
    query: sum(rate(http_requests_total[5m])) by (service)
    
  - title: "Error Rate"
    type: stat
    query: |
      sum(rate(http_requests_total{status=~"5.."}[5m]))
      / sum(rate(http_requests_total[5m])) * 100
    thresholds:
      - color: green
        value: 0
      - color: red
        value: 1
        
  - title: "P99 Latency"
    type: gauge
    query: histogram_quantile(0.99, sum(rate(http_request_duration_seconds_bucket[5m])) by (le))
    
  - title: "Top Endpoints"
    type: table
    query: topk(10, sum(rate(http_requests_total[5m])) by (endpoint))

7.2 Dependency Map

---

8 - Alerting APM

# alerts.yml
groups:
  - name: apm-alerts
    rules:
      - alert: HighLatencyP99
        expr: |
          histogram_quantile(0.99, 
            sum(rate(http_request_duration_seconds_bucket[5m])) by (service, le)
          ) > 1
        for: 5m
        labels:
          severity: warning
        annotations:
          summary: "High P99 latency on {{ $labels.service }}"
          
      - alert: ApdexLow
        expr: |
          (
            sum(rate(http_request_duration_seconds_bucket{le="0.5"}[5m])) by (service)
            + sum(rate(http_request_duration_seconds_bucket{le="2.0"}[5m])) by (service) / 2
          )
          / sum(rate(http_request_duration_seconds_count[5m])) by (service)
          < 0.8
        for: 10m
        labels:
          severity: critical

---

Résumé

Dans ce chapitre, nous avons appris :

• Les concepts APM
• Les solutions disponibles
• Elastic APM et agents
• SigNoz open source
• Le Profiling continu
• Le RUM (Real User Monitoring)
• Les Dashboards et alertes APM

---

Prochaine étape

Dans le prochain chapitre, nous verrons les Bonnes pratiques.

→ Chapitre suivant : Bonnes pratiques

---

← Retour à la table des matières