Skip to content

Service Comparison - Observability and Monitoring

Overview

This guide compares monitoring, logging, tracing, and alerting services across AWS, Azure, and Google Cloud. Observability is a foundational topic for cloud operations, SRE, and DevOps certification exams.

Core Monitoring Services

CloudWatch vs Azure Monitor vs Cloud Monitoring

Feature AWS CloudWatch Azure Monitor Google Cloud Monitoring
Metrics Collection Automatic (AWS services) Automatic (Azure services) Automatic (GCP services)
Custom Metrics CloudWatch PutMetricData Custom metrics API Custom metrics / OpenTelemetry
Metric Resolution 1-second (high-res), 60-second (standard) 1-minute (standard) 10-second (custom), 60-second (default)
Metric Retention 15 months (varies by resolution) 93 days (platform), configurable 24 months
Dashboards CloudWatch Dashboards Azure Dashboards + Workbooks Cloud Monitoring Dashboards
Cross-Account CloudWatch cross-account observability Azure Lighthouse Multi-project monitoring
Anomaly Detection CloudWatch Anomaly Detection (ML) Smart detection (App Insights) N/A (use MQL queries)
Composite Alarms Yes N/A Yes (MQL alert policies)
Pricing Per metric, per API call, per dashboard Per GB ingested, per alert rule Per MB ingested (free allotment)
Free Tier 10 custom metrics, 10 alarms, 3 dashboards First 5 GB/month First 150 MB/month + free allotment

Metric Types and Namespaces

Aspect AWS CloudWatch Azure Monitor Google Cloud Monitoring
Namespace Format AWS/ServiceName Microsoft.ServiceName compute.googleapis.com
Dimensions Up to 30 per metric Dimension filtering Labels (up to 30)
Statistics Sum, Average, Min, Max, p50-p99 Aggregations (Avg, Sum, Min, Max, Count) Aligners and reducers
Query Language CloudWatch Metrics Insights Kusto Query Language (KQL) Monitoring Query Language (MQL)
Math Expressions Metric math KQL functions MQL functions
Cross-Service Queries Yes (Metrics Insights) Yes (KQL) Yes (MQL)

Logging Services

CloudWatch Logs vs Azure Monitor Logs vs Cloud Logging

Feature AWS CloudWatch Logs Azure Monitor Logs Google Cloud Logging
Log Ingestion CloudWatch agent, SDKs, API Azure Monitor agent, SDKs Logging agent, SDKs, API
Structured Logging JSON (Embedded Metric Format) Structured (KQL parsing) JSON (structured payloads)
Query Language CloudWatch Logs Insights Kusto Query Language (KQL) Cloud Logging filter syntax
Log Groups/Tables Log groups + log streams Log Analytics workspaces + tables Log buckets + log views
Retention 1 day to 10 years (configurable) 30-730 days (workspace), up to 12 years (archive) 1-3,650 days (configurable)
Archive Storage S3 export Archive tier (Blob Storage) Cloud Storage export
Real-time Streaming Kinesis Data Firehose, Lambda Event Hubs Pub/Sub
Log-based Metrics Metric filters KQL-based alerts Log-based metrics
Cross-Account Cross-account log sharing Cross-workspace queries Cross-project log sinks
Pricing Ingestion + storage + queries Ingestion + retention Ingestion + storage (free 50 GB/month)
Free Tier 5 GB ingestion, 5 GB storage 5 GB ingestion/month 50 GB ingestion/month

Log Collection Agents

Feature AWS Azure GCP
Primary Agent CloudWatch unified agent Azure Monitor Agent (AMA) Ops Agent
Legacy Agent CloudWatch Logs agent Log Analytics agent (MMA) - deprecated Legacy Logging agent
Container Logging Fluent Bit (FireLens) Container Insights (AMA) Fluent Bit (GKE)
Configuration SSM Parameter Store / JSON config Data Collection Rules (DCR) YAML configuration
Multi-Destination Limited Multiple workspaces Multiple sinks
OpenTelemetry ADOT Collector Azure Monitor OpenTelemetry Google Cloud OpenTelemetry

Distributed Tracing

X-Ray vs Application Insights vs Cloud Trace

Feature AWS X-Ray Azure Application Insights Google Cloud Trace
Type Distributed tracing APM + distributed tracing Distributed tracing
Auto-Instrumentation X-Ray SDK, ADOT Auto-instrumentation (codeless) Auto-instrumentation
Language Support Java, Python, Node.js, .NET, Go, Ruby Java, Python, Node.js, .NET, JavaScript Java, Python, Node.js, Go, .NET, Ruby
OpenTelemetry ADOT (AWS Distro for OpenTelemetry) Native OpenTelemetry support Native OpenTelemetry support
Sampling Fixed rate, reservoir Adaptive sampling Automatic / fixed rate
Service Map X-Ray service map Application Map Trace overview
Trace Analysis Trace analytics (filter/group) Transaction diagnostics Trace list and details
Latency Analysis Response time distribution Performance blade Latency analysis
Error Analysis Error/fault tracking Failure analysis Error reporting (separate)
Correlation Trace ID propagation Operation ID / W3C trace context W3C trace context
Pricing Per trace recorded + scanned Per GB ingested $0.20/million spans
Free Tier 100,000 traces/month 5 GB/month First 2.5 million spans/month

Application Performance Monitoring (APM)

Full APM Comparison

Feature AWS (X-Ray + CloudWatch) Azure Application Insights Google Cloud (Trace + Error Reporting)
End-to-End APM Partial (multiple services) Yes (unified) Partial (multiple services)
Real User Monitoring CloudWatch RUM Browser SDK (RUM) N/A (use third-party)
Synthetic Monitoring CloudWatch Synthetics Availability tests Uptime checks
Profiling CodeGuru Profiler Application Insights Profiler Cloud Profiler
Dependency Tracking X-Ray subsegments Auto-dependency tracking Trace spans
Live Metrics N/A Live Metrics Stream N/A
Smart Detection Anomaly Detection Smart Detection alerts N/A
Snapshot Debugging N/A Snapshot Debugger Cloud Debugger (deprecated)

Alerting and Notification

Alert Services Comparison

Feature AWS CloudWatch Alarms Azure Monitor Alerts Google Cloud Alerting
Metric Alerts Standard + Composite Metric alerts Metric-based policies
Log Alerts Metric filters + alarms Log search alerts (KQL) Log-based alerts
Anomaly Alerts Anomaly detection band Dynamic thresholds N/A (use MQL)
Multi-Condition Composite alarms Multi-resource alerts Multi-condition policies
Action Types SNS, Lambda, EC2, Auto Scaling Action groups (email, SMS, webhook, Logic Apps, Functions) Notification channels (email, SMS, Slack, PagerDuty, webhooks)
Severity Levels OK, ALARM, INSUFFICIENT_DATA Sev 0-4 No severity (critical/warning via policy)
Alert Suppression N/A Alert processing rules Snooze
Pricing $0.10/alarm/month (standard) Per alert rule/month Per policy/month
Evaluation Frequency 10-second to 1-day 1-minute to 1-day 30-second to 1-day

Notification Services

Feature AWS SNS Azure Action Groups Google Notification Channels
Email Yes Yes Yes
SMS Yes Yes Yes
Voice Call No Yes No
Webhook Yes (HTTP/S) Yes Yes
Slack Via AWS Chatbot Via Logic Apps Yes (native)
PagerDuty Via SNS subscription Via Action Groups Yes (native)
Teams Via AWS Chatbot Yes (native) Via webhook
OpsGenie Via SNS subscription Via Action Groups Yes (native)
Mobile Push Yes (SNS push) Azure mobile app Google Cloud mobile app

Managed Prometheus and Grafana

Feature Amazon Managed Prometheus Azure Monitor (Prometheus) Google Managed Prometheus
Collection ADOT, Prometheus remote write Azure Monitor agent (Prometheus) Managed collection, self-deploy
Query PromQL PromQL (Azure Monitor Workspace) PromQL (Cloud Monitoring)
Storage Managed (AMP workspace) Azure Monitor Workspace Cloud Monitoring metrics store
Retention 150 days 18 months 24 months
Grafana Integration Amazon Managed Grafana Azure Managed Grafana Self-hosted / Cloud Monitoring
AlertManager Managed AlertManager Prometheus alert rules N/A (use Cloud Alerting)
Pricing Per metric sample ingested + stored + queried Per metrics ingested Per metric sample ingested

Infrastructure Monitoring

VM/Instance Monitoring

Feature AWS Azure GCP
Basic Metrics CloudWatch (5-min default) Azure Monitor (1-min) Cloud Monitoring (1-min)
Detailed Metrics Detailed monitoring (1-min, paid) VM Insights Agent metrics
Process Monitoring CloudWatch agent (procstat) VM Insights (Map) Ops Agent (process plugin)
Agent CloudWatch unified agent Azure Monitor Agent Ops Agent
Guest OS Metrics Agent-based Agent-based Agent-based

Container Monitoring

Feature AWS Azure GCP
Service Container Insights Container Insights GKE Monitoring
Metrics Node, pod, container, service Node, pod, container Node, pod, container, workload
Logs CloudWatch Logs (Fluent Bit) Log Analytics (ContainerLogV2) Cloud Logging (Fluent Bit)
Prometheus Amazon Managed Prometheus Azure Monitor (Prometheus) Google Managed Prometheus
Cost Visibility Kubecost integration AKS Cost Analysis GKE Cost Allocation
Control Plane Logging EKS control plane logging AKS diagnostics GKE system logs

Database Monitoring

Feature AWS Azure GCP
Performance Insights RDS Performance Insights Intelligent Insights Query Insights
Slow Query Logs CloudWatch Logs integration Azure Monitor Logs Cloud Logging
Metrics CloudWatch RDS metrics Azure Monitor metrics Cloud Monitoring metrics
Recommendations Trusted Advisor Azure Advisor Recommender

Network Monitoring

Feature AWS Azure GCP
Flow Logs VPC Flow Logs NSG Flow Logs VPC Flow Logs
Traffic Analysis Traffic Mirroring Traffic Analytics Packet Mirroring
Network Monitor CloudWatch Network Monitor Network Watcher Network Intelligence Center
Connectivity Testing Reachability Analyzer Connection Monitor Connectivity Tests
Latency Monitoring CloudWatch Internet Monitor Connection Monitor Network Intelligence Center
DNS Logging Route 53 query logs DNS Analytics Cloud DNS query logs

Synthetic Monitoring

Feature AWS CloudWatch Synthetics Azure Availability Tests Google Uptime Checks
HTTP Checks Yes Yes (URL ping, Standard) Yes
Browser Tests Canary scripts (Node.js) Multi-step web tests N/A
Custom Scripts Puppeteer-based N/A N/A
Global Locations 20+ regions 16+ locations 20+ regions
SSL Monitoring Yes Yes Yes
Screenshot Capture Yes N/A N/A
Frequency 1-minute minimum 5-minute minimum 1-minute minimum
Pricing Per canary run Included / per web test Free (basic), paid (advanced)

OpenTelemetry Integration

Feature AWS Azure GCP
Distribution AWS Distro for OpenTelemetry (ADOT) Azure Monitor OpenTelemetry Distro Google Cloud OpenTelemetry
Auto-Instrumentation ADOT auto-instrumentation Auto-instrumentation (Java, .NET, Node.js, Python) Auto-instrumentation
Collector ADOT Collector Azure Monitor Exporter Google Cloud Exporter
Export Targets X-Ray, CloudWatch, AMP Application Insights, Azure Monitor Cloud Trace, Cloud Monitoring
Kubernetes ADOT add-on (EKS) AKS OpenTelemetry GKE auto-instrumentation

Certification Exam Focus Areas

AWS SysOps Administrator / DevOps Professional

  • CloudWatch Logs Insights query syntax
  • CloudWatch agent configuration and custom metrics
  • X-Ray trace analysis and service maps
  • CloudWatch Synthetics canary scripts
  • Composite alarms and anomaly detection
  • Container Insights setup and metrics

Azure Administrator / Monitor Workbook Author

  • KQL query syntax for Log Analytics
  • Azure Monitor agent vs legacy agents
  • Application Insights configuration modes
  • Alert processing rules and action groups
  • VM Insights and dependency mapping
  • Workbook authoring and visualization

Google Cloud Operations / SRE

  • Cloud Logging filter syntax and log-based metrics
  • MQL (Monitoring Query Language) basics
  • Uptime checks and alerting policies
  • Cloud Trace integration with OpenTelemetry
  • Error Reporting grouping and notifications
  • SLO monitoring and error budgets
  • AWS CloudWatch: https://docs.aws.amazon.com/cloudwatch/
  • AWS X-Ray: https://docs.aws.amazon.com/xray/latest/devguide/
  • Azure Monitor: https://learn.microsoft.com/en-us/azure/azure-monitor/
  • Azure Application Insights: https://learn.microsoft.com/en-us/azure/azure-monitor/app/app-insights-overview
  • Google Cloud Monitoring: https://cloud.google.com/monitoring/docs
  • Google Cloud Logging: https://cloud.google.com/logging/docs
  • Google Cloud Trace: https://cloud.google.com/trace/docs
  • OpenTelemetry: https://opentelemetry.io/docs/

Key Takeaways

  1. Azure Monitor with Application Insights provides the most comprehensive single-platform APM experience
  2. CloudWatch offers the deepest AWS integration but requires combining multiple services for full observability
  3. Google Cloud provides the most generous free tier for logging (50 GB/month)
  4. All three providers are converging on OpenTelemetry as the standard instrumentation framework
  5. Managed Prometheus is available on all three platforms - PromQL skills are highly transferable
  6. KQL (Azure) is the most powerful query language but has the steepest learning curve
  7. For multi-cloud observability, consider third-party tools like Datadog, New Relic, or Grafana Cloud
  8. Understanding log retention policies and archive strategies is critical for compliance exam questions

Related Guides: - LLM Observability Comparison - DevOps and CI/CD Comparison - Containers and Kubernetes Comparison - Topic: Observability - Observability Basics