GCP Professional Cloud Architect Practice Question

A financial-services firm runs a stateful trading platform on Compute Engine virtual machines in a regional managed instance group that spans two zones in us-central1. Each VM stores its trading logs on a regional Persistent Disk. Corporate resilience policy stipulates that, if the entire region fails, the application must be restored in any other US region within one hour with no more than four hours of data loss. Operations wants the backup solution to be largely self-managing, cost-efficient, and compatible with customer-managed encryption keys (CMEK). Which approach best meets these requirements?

Create a Nearline Cloud Storage bucket in us-central1 and schedule a cron job on each VM to rsync disk contents to the bucket every hour; during failover, copy objects back to new disks before starting the application.
Attach a second Persistent Disk in another region over Cloud Interconnect and configure synchronous block-level replication between the two disks; rely on the managed instance group to restart VMs in that region with zero data loss.
Create an hourly snapshot schedule on each regional Persistent Disk, ensure the disks are CMEK-encrypted so snapshots inherit the key, and set the snapshot storage location to the multi-region "us"; during a regional outage, create new disks from the latest snapshot in the target US region and recreate the managed instance group.
Provision a Filestore Enterprise instance with cross-region replication and mount it to all VMs; during a regional outage, recreate the VMs in another region and remount the replicated Filestore volume.

Report Issue

Answer Description

Creating hourly scheduled snapshots of the regional Persistent Disks delivers a worst-case recovery-point objective of one hour, which satisfies the requirement of no more than four hours of data loss. Storing the snapshots in the multi-region "us" location automatically replicates them to at least two U.S. regions, ensuring availability even if the entire us-central1 region is unavailable. Because Persistent Disk snapshots are global resources, you can quickly create new disks from the latest snapshot in any U.S. region and attach them to replacement instances, meeting the one-hour recovery-time objective. Snapshot schedules run without manual intervention, and snapshots of CMEK-encrypted disks inherit the same key, maintaining encryption compliance while keeping storage costs low through incremental, compressed storage.

The Filestore alternative adds higher ongoing storage and replication costs and still requires manual VM recreation. The rsync solution lacks crash consistency and stores data in a single region until restore. Persistent Disk does not support synchronous cross-region block replication, so attaching a secondary disk in another region over Cloud Interconnect is not feasible.

Ask Bash

Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.