Associate Cloud Engineer ยท Choosing and deploying data services
Multi-region redundancy for data
Multi-region and dual-region options, cross-region replicas, Spanner multi-region, BigQuery dataset location, RPO/RTO trade-offs.
Key points
- Cloud SQL HA: covers a zone failure.
- Cross-region read replica: can be promoted after a regional outage (RPO depends on replication lag, RTO on promoting and repointing clients).
- Spanner: regional 99.99%; dual-region and multi-region 99.999% (Enterprise Plus), read-write plus witness replicas.
- BigQuery: dataset location fixed at creation; a multi-region (US, EU) alone gives no cross-region redundancy; use managed disaster recovery.
Source: https://docs.cloud.google.com/bigquery/docs/datasets
Practice questions
When can the location of a BigQuery dataset be set?
- Only when the dataset is created
- At any time with bq update
- Only while the dataset is empty
- Whenever a table is added
Show answer
Answer: A. Only when the dataset is created
The location is chosen at creation and cannot be changed afterwards; data would have to be copied to a new dataset in the other location.
Which statements about Spanner instance configurations are correct?
- Multi-region configurations use two read-write regions plus a witness region.
- Witness replicas store a full copy of the data and serve reads.
- Dual-region configurations keep three replicas in each of two regions.
- Read-only replicas take part in write quorums.
- A regional configuration offers a 99.99% availability SLA.
Show answer
Answer: A. Multi-region configurations use two read-write regions plus a witness region., C. Dual-region configurations keep three replicas in each of two regions., E. A regional configuration offers a 99.99% availability SLA.
Multi-region configs have two read-write regions plus a witness region; dual-region configs hold six replicas, three per region; regional configs offer 99.99%. Witness replicas hold no data, and read-only replicas do not join write quorums.
True or false: choosing the US multi-region for a BigQuery dataset automatically gives you cross-region replication.
Show answer
Answer: False
False. A multi-region is a location that stores data within one region chosen by BigQuery; cross-region redundancy needs managed disaster recovery.
Practice 8 questions on this topic in Bitzu
Short lessons, instant explanations and spaced review of what you get wrong. Free on iOS and Android.
