Intrigued by the art of cloud architecture? Discover how to design, develop, and manage robust, secure, scalable, and dynamic solutions on Google Cloud as you prepare for the Professional Cloud Architect exam!
Gauge your current knowledge
Gauge your current knowledge
A Business Continuity Plan starts by defining two key numbers: the Recovery Time Objective (RTO) and the Recovery Point Objective (RPO). The RTO is the maximum acceptable downtime after a failure, while the RPO is the maximum acceptable data loss, measured in time. These goals determine how quickly and completely your system must recover, guiding your choice of architectural pattern.
A Cold Standby pattern is the most cost-effective option. Key resources, like virtual machines, are not running until a disaster occurs. Recovery involves creating new infrastructure from saved snapshots or images. This pattern results in a higher RTO because starting everything from scratch takes time, and it often requires manual steps to begin the recovery process.
A Warm Standby pattern keeps a scaled-down, always-running version of your system in a secondary region. This "pilot light" approach means core services are active and ready. If the primary region fails, the standby system can quickly scale up to full capacity using tools like Managed Instance Groups. This offers a faster recovery than cold standby, balancing cost and readiness.
A Hot Standby (or active-active) pattern provides the highest resilience. Systems run simultaneously in multiple regions, with Cloud Load Balancing distributing live user traffic across them all. Data is replicated synchronously, so there is no data loss (low RPO) and traffic can shift instantly if one region fails (low RTO). This pattern is designed for mission-critical applications requiring maximum uptime.
Choosing the right pattern is a balance between the cost of the solution and the financial impact of potential downtime. Architects use Regional and Multi-region Resources to protect against local failures. It is critical to regularly test your disaster recovery plan to ensure your chosen pattern will work as expected during a real emergency.
Data Resiliency ensures data survives a regional disaster. Managed services like Cloud Storage, Firestore, and Spanner offer multi-regional options that automatically distribute and replicate your data across geographically separate areas. This provides high availability without you managing the underlying replication.
Cloud Storage achieves extreme durability. For the highest protection, you can configure buckets as Dual-region or Multi-region, which store copies in at least two separate geographic locations. Features like turbo replication can guarantee a specific Recovery Point Objective (RPO), such as 15 minutes, meaning your data loss in a disaster is limited to that time window.
Managed databases have built-in features for resilience. Cloud SQL can use Cross-region Replicas that can be promoted to primary if the main region fails. Spanner and Bigtable also provide automatic replication across zones and regions. These features maintain data synchronization and enable quick failover for operational continuity.
To protect against accidental deletion or corruption, automated backups are essential. Point-in-Time Recovery (PITR) allows you to restore a database to its exact state at any minute within a recent window (e.g., the last 7 days). Scheduled snapshots provide longer-term backups. Retention policies control how long this backup data is kept to meet business or legal requirements.
A complete strategy must balance Recovery Time Objective (RTO) and Recovery Point Objective (RPO). While multi-regional storage protects data, compute resources like Compute Engine instances must also be configured to failover. Active-active architectures, where workloads run in parallel in two locations, offer the highest reliability by ensuring both data and application processing are always available somewhere.
Traffic management and automated failover are critical for meeting RTO and RPO goals by minimizing manual recovery steps. Google Cloud services work together to automatically redirect user traffic away from failed infrastructure, keeping services running with little or no interruption.
Global Cloud Load Balancing is the primary tool for directing user traffic. It uses continuous health checks to monitor the status of backend servers. If these checks fail for a zone or region, the load balancer automatically reroutes traffic to healthy instances in another location, ensuring users always reach a working endpoint.
Cloud DNS adds another layer of control with configurable routing policies. These policies can be set up for failover routing, which automatically switches users to a backup IP address if the primary destination is unhealthy. Other policies, like geolocation routing, direct users to resources based on their physical location for performance or compliance reasons.
Managed Instance Groups (MIGs) maintain a set of identical virtual machines across zones. If an instance fails, the MIG automatically recreates it using a predefined instance template. This autohealing capability, combined with regional persistent disks that keep data available across zones, provides a robust foundation for recovering from zonal failures.
For protection against a disaster affecting an entire region, a multi-region architecture is necessary. This involves replicating data and deploying application resources across multiple global regions. By spreading the workload this way, organizations can achieve very high availability, ensuring that a failure in one geographic area does not cause a total service outage.
StudyCerts
StudyCerts