Question 800
The previous outage involved this problem: A Bigtable replicated workload routes consistency-sensitive read-after-write operations to a different cluster immediately after updates. What long-term control addresses the risk?
Explanation
Correct Answer
Document replication consistency expectations for client routing profiles. The recurring risk is multi-cluster replication freshness was mistaken for synchronous linearizable read behavior. This check makes that risk visible before another incident.
Incorrect Answers
This control would include schema and IAM smoke tests after Spanner infrastructure provisioning; it responds to the risk in which a Spanner application fails to read a new table after deployment despite successful instance creation. It would not address the current root cause: multi-cluster replication freshness was mistaken for synchronous linearizable read behavior.
This control would treat schema backfills as capacity-affecting rollout events; it responds to the risk in which a Spanner schema change introduces a new index and large backfill work without checking user-facing transaction latency. It would not address the current root cause: multi-cluster replication freshness was mistaken for synchronous linearizable read behavior.
The proposed check would use recoverability checks as a mandatory step of database provisioning, helping with cases where a new database instance is marked production ready because it accepts logins, but no automated backup has ever completed. The recurring exposure described here is that multi-cluster replication freshness was mistaken for synchronous linearizable read behavior.