Monitoring, Logging, and Cost Management

Understanding how to monitor your GCP resources and manage costs is just as important as knowing how to build with them. GCP provides comprehensive observability tools through Google Cloud Observability (formerly Stackdriver) and robust billing management features.

---

Cloud Monitoring

Cloud Monitoring collects metrics, creates dashboards, and triggers alerts across your GCP resources and applications.

Metrics

Cloud Monitoring automatically collects metrics from GCP services:

Service	Example Metrics
Compute Engine	CPU utilisation, disk I/O, network traffic
Cloud SQL	Database connections, query latency, storage usage
Cloud Run	Request count, request latency, instance count
Cloud Functions	Invocations, execution time, errors
Cloud Storage	Total bytes, object count, request count

Custom Metrics

You can create your own metrics using the Cloud Monitoring API or OpenTelemetry:

from google.cloud import monitoring_v3

client = monitoring_v3.MetricServiceClient()
# Write custom metric data points

Dashboards

Cloud Monitoring provides:

• Pre-built dashboards for GCP services (Compute Engine, GKE, Cloud SQL, etc.)
• Custom dashboards where you choose which metrics and charts to display
• Widgets — line charts, bar charts, gauges, scorecards, and logs panels

---

Alerting

Alerts notify you when metrics cross a threshold or when specific conditions are met.

Alert Policy Components

Component	Description
Condition	The metric threshold or absence condition to monitor
Duration	How long the condition must be met before firing
Notification channel	Where to send the alert (email, SMS, Slack, PagerDuty, webhook)
Documentation	Instructions for responders included in the alert

Example Alert

"Alert me if Compute Engine CPU utilisation exceeds 80% for 5 minutes."

---

Cloud Logging

Cloud Logging is a centralised log management service for collecting, storing, searching, and analysing logs.

Log Types

Type	Description
Platform logs	Generated by GCP services (Cloud SQL, GKE, Cloud Run, etc.)
User logs	Sent from your applications via client libraries or agents
Audit logs	Record administrative actions (Admin Activity, Data Access, System Event)
Network logs	VPC Flow Logs, firewall rule logs

Log Explorer

The Log Explorer in the Console lets you:

• Search and filter logs across all projects and services
• Build queries using the logging query language
• Create metrics from log entries (logs-based metrics)
• Stream logs in real time

Example Query

resource.type="cloud_run_revision"
severity>=ERROR
textPayload:"connection refused"

Log Retention

Log Type	Default Retention
Admin Activity	400 days (no charge)
Data Access	30 days
Platform / User logs	30 days

For longer retention, route logs to Cloud Storage or BigQuery.

---

Cloud Audit Logs

Audit logs answer the questions: Who did what, where, and when?

Audit Log Type	What It Records
Admin Activity	Configuration changes (create/delete/update resources) — always enabled
Data Access	Read operations on resource data (optional, can be verbose)
System Event	Google-initiated administrative actions (live migration, auto-scaling)
Policy Denied	Actions denied by IAM or organisation policies

---

Error Reporting

Error Reporting automatically groups and counts application errors: