Multicloud Operations, Observability, and Cost Management
Operating model
Multicloud operations need one service model, shared runbooks, and clear ownership. Define who patches what: Oracle manages the database platform in Database@ partner clouds; you manage data, schemas, access policies, and application configuration. Use consistent tagging such as cost center, application, environment, and owner across OCI, Azure, AWS, and Google Cloud.
Observability
- OCI Observability provides metrics, logs, alarms, and tracing for OCI resources; partner clouds use Azure Monitor, CloudWatch, and Google Cloud Monitoring.
- Export database and audit logs to a central SIEM or log archive; correlate by request ID and timestamp.
- Monitor private connectivity, BGP state, latency, packet loss, and database wait events.
- Alert on backup failures, Data Guard lag, storage headroom, and identity federation health.
Cost management
Use OCI Cost Analysis and budgets, Azure Cost Management, AWS Cost Explorer, and Google Cloud Billing. Watch data egress, interconnect ports, cross-region replication, and idle Exadata capacity. BYOL and Universal Credits can change economics; marketplace purchases may count toward partner commitments. Rightsize and schedule nonproduction databases.
Worked example
A team runs Autonomous Database@Azure for production and a small OCI dev database. They tag both app=orders. Azure Cost Management shows marketplace spend; OCI shows dev compute. A dashboard tracks GoldenGate lag, interconnect latency, and monthly egress. The architect removes an idle cross-cloud replication link and moves backups to lower-cost object storage, cutting spend without reducing RTO.
Traps
Monitoring only one cloud; alerting on infrastructure but not replication lag; leaving default tags; ignoring egress and interconnect port charges; assuming marketplace spend is invisible to OCI budgets; over-provisioning standby capacity.