Skip to main content

Operations

Exam guide§3.4, 2.4

Day-2: seeing what your resources are doing, and provisioning them repeatably.

Why observability​

On-premises you can physically touch the servers - if an application hangs, someone can walk over and find out why. In the cloud the servers are Google's, not yours, and you can't inspect them directly. So the only way to know what's happening with your server, database, or application is Google's integrated observability tools.

On-premisesthe servers are yourswalk over and inspect them✓ you can look insideCloudthe servers are Google'syou can't touch them - a black box✗ no way to look inside directlyyou already have visibility -nothing extra needed to see your own boxescan't see in, so you needGoogle Cloud Observability toolscapture signals - metrics, logs, traces -the only way to see inside the cloud black box
On-premises you can physically inspect a failing server. In the cloud the servers are Google's and opaque to you, so integrated observability tools are the only way to see what your systems are doing.

Users come to observability with four recurring needs:

Visibility into system healthA clear model of how the app runs:"are systems functioning? enough resources?"Error reporting and alertingHealth at a glance, plus proactivealerting and anomaly detectionEfficient troubleshootingCorrelate logs and metrics in one place,with a clear action to takePerformance improvementAnalyze trends to plan ahead and seehow changes affect performance
Four recurring needs drive observability: know your systems are healthy, be told when they are not, troubleshoot fast, and improve performance over time.

Google Cloud Observability​

Google Cloud Observability (formerly the operations suite / Stackdriver) is one integrated service for monitoring, logging, and diagnostics - not a loose collection of separate packages. It dynamically discovers your cloud resources and application services through deep integration with Google Cloud and Amazon Web Services, so its smart defaults give you core visibility across your platform in minutes.

On top of that discovery you get powerful data and analytics tools, open-source agents and integrations, and collaboration with many third-party software providers.

Google Cloud ObservabilityMonitoringLoggingError ReportingTraceProfiler
One integrated service fans out to five products - Monitoring, Logging, Error Reporting, Trace, and Profiler - so telemetry, alerting, and diagnostics work together instead of as loosely coupled tools.
GotchaOne integrated service, not separate packages

In most other environments monitoring, logging, error reporting, and tracing are handled by completely different packages or a loosely integrated collection of software. Google Cloud Observability ships them as a single, comprehensive, integrated service - that cohesion is what makes applications reliable, stable, and maintainable.

Signals and open standards​

Observability rests on three signals - metrics, logs, and traces - captured through one open, vendor-neutral instrumentation layer. Metrics detect that something is wrong, logs diagnose what happened, and traces show where and how:

Observabilityunderstand how the system behavesMetricsDetects a problemnumeric measurements over timeCPU %, request rate, errorsPrometheus · PromQLLogsDiagnoses what happenedtimestamped event recordserrors, audit + app messagesFluentbit · Ops AgentTracesShows where and howa request's path across serviceslatency per service hopOpenTelemetry SDKOpenTelemetryone vendor-neutral instrumentation layer - a trace ID links a log line to the request that produced it
The three signals of observability sit on one open instrumentation layer: metrics detect a problem, logs diagnose what happened, traces show where and how. OpenTelemetry unifies them so a single trace ID can link a log line to the request that produced it.

The observability workflow​

Signals feed the tools, the tools drive incident management, and it all points at faster troubleshooting:

Capture SignalsMetricsLogsTraceVisualize and AnalyzeDashboardsMetrics ExplorerLogs ExplorerService MonitoringLog AnalyticsHealth ChecksManage IncidentsAlertsError ReportingSLOTroubleshoot
The observability workflow: capture signals (metrics, logs, traces), visualize and analyze them, manage incidents when they fire, and drive it all to troubleshooting.
FactsPricing model
  • You pay only for what you use - no upfront fees or commitments.
  • Free usage allotments let you get started at no cost.
  • For current pricing, refer to the Google Cloud Observability pricing docs.
0%0 of 147 pages studied