Regions, dual-regions, multi-regions and zones for redundancy
Lesson coming
Sample questions
Try each one before opening the answer. Every option is explained, with the Google Cloud documentation page that proves it.
Question 1 · choose 1
A Cloud Storage bucket in the us-east1 region holds files that dozens of jobs and partner integrations address by the bucket's name. A new resilience rule says the data must survive the loss of a whole region and must be stored only in us-east1 and us-east4. The team wants to keep the bucket and its name, reads must continue during the change, and a short pause in writes is acceptable. Storage Intelligence is configured for the project. What should the team do?
ARelocate the bucket to a configurable dual-region of us-east1 and us-east4
BEnable turbo replication on the bucket with us-east4 as the second region
CChange the bucket's default storage class to Multi-Regional storage
DRelocate the bucket, keeping its name, to the predefined dual-region NAM4 in the US
Show the answer and why
ARelocate the bucket to a configurable dual-region of us-east1 and us-east4
Correct
Bucket relocation moves an existing bucket to another location, such as from a region to a dual-region, without changing its name; reads continue, and writes pause only during the final synchronization.
BEnable turbo replication on the bucket with us-east4 as the second region
Incorrect
Turbo replication is only available for buckets in dual-regions; a regional bucket has no second region to replicate to.
CChange the bucket's default storage class to Multi-Regional storage
Incorrect
Multi-Regional is a legacy storage class that can only be used for objects in multi-regions or dual-regions; a storage class does not change where a regional bucket keeps its data.
DRelocate the bucket, keeping its name, to the predefined dual-region NAM4 in the US
Incorrect
NAM4 is the predefined pair of us-central1 (Iowa) and us-east1 (South Carolina), so part of the data would be stored outside us-east1 and us-east4.
A bucket's location type sets its redundancy: a regional bucket keeps data in one region, while a dual-region is a specific pair of regions (a predefined pair such as NAM4, or a configurable pair you choose) and a multi-region spans a large area. With Storage Intelligence configured, bucket relocation moves an existing bucket to a new location, for example from a single region to a dual-region for resilience, keeping the bucket name and object metadata. Reads continue throughout; relocations with write downtime lock writes only during the final synchronization step. Turbo replication is a setting of dual-region buckets.
A company stores reporting extracts in Cloud Storage. Its regulator requires that copies exist in exactly two named regions, Frankfurt and the Netherlands, so that a regional outage does not stop access. The recovery point objective is 12 hours, and the company wants to pay no more than this requires. Which bucket setup fits?
AA dual-region bucket in Frankfurt and the Netherlands with turbo replication turned on
BA regional bucket in Frankfurt, copied every hour to a regional bucket in the Netherlands by a Storage Transfer Service job
CA bucket in the predefined EUR4 dual-region with default replication
DA dual-region bucket in Frankfurt and the Netherlands with default replication
Show the answer and why
AA dual-region bucket in Frankfurt and the Netherlands with turbo replication turned on
Incorrect
Turbo replication is a premium feature that targets 15 minutes; it meets the RPO but costs more than a 12-hour RPO needs.
BA regional bucket in Frankfurt, copied every hour to a regional bucket in the Netherlands by a Storage Transfer Service job
Incorrect
The hourly copy meets the RPO, but a regional bucket fails over only between zones; during a Frankfurt outage applications would have to change storage paths to the other bucket before access resumed.
CA bucket in the predefined EUR4 dual-region with default replication
Incorrect
EUR4 pairs europe-north1 (Finland) with europe-west4 (Netherlands), so it does not keep a copy in Frankfurt.
DA dual-region bucket in Frankfurt and the Netherlands with default replication
Correct
Both regions use the EU location code, so they can form a configurable dual-region that fails over automatically; default replication targets 100% of new objects across regions within 12 hours, matching the RPO without the turbo premium.
Dual-regions give cross-region redundancy with precise control of the two regions and automatic failover: predefined pairs such as EUR4 (Finland and Netherlands) or configurable pairs of regions with the same location code. Default replication targets 99.9% of new objects within one hour and 100% within 12 hours; turbo replication, a premium option for dual-regions, targets 15 minutes.
A BigQuery table was deleted 12 days ago in a dataset whose time travel window is the default seven days, and no snapshot or backup of it exists. The team now needs the table back. What is the remaining option?
ARaise the dataset's time travel window to 14 days, then query the table as of a time before the deletion
BNone; fail-safe keeps deleted data for seven days after the deletion, and that period has passed
CRestore the table yourself with a copy job that uses a timestamp decorator from just before the deletion
DAsk Cloud Customer Care for fail-safe recovery access, then copy the table with a timestamp decorator
Show the answer and why
ARaise the dataset's time travel window to 14 days, then query the table as of a time before the deletion
Incorrect
The time travel window can be set only from two to seven days, so it can't be raised to 14 days, and time travel cannot reach 12 days back.
BNone; fail-safe keeps deleted data for seven days after the deletion, and that period has passed
Incorrect
Fail-safe starts when the time travel window ends and lasts seven more days, so data deleted 12 days ago is still within the 14 days covered.
CRestore the table yourself with a copy job that uses a timestamp decorator from just before the deletion
Incorrect
A copy job you start on your own reaches only into the time travel window, seven days here; a table can be copied from fail-safe storage only after Cloud Customer Care grants temporary recovery access.
DAsk Cloud Customer Care for fail-safe recovery access, then copy the table with a timestamp decorator
Correct
Deleted data stays in fail-safe for seven more days after the time travel window, so 12 days after deletion it is still there. Customer Care grants temporary recovery access, and you restore the table with a copy job.
BigQuery keeps changed and deleted data for the time travel window (two to seven days, seven by default), which you can query or copy yourself, and then for a seven-day fail-safe period. Fail-safe data cannot be queried directly; Cloud Customer Care must grant recovery access, and after the fail-safe period nothing can be recovered.
A manager asks whether choosing Archive storage or a single-region bucket lowers the annual durability that Cloud Storage is designed for. What is the answer?
ANo; Cloud Storage is designed for at least 11 nines of annual durability whatever the storage class and location type
BYes; colder classes are designed for lower durability in exchange for their lower price
CYes; only multi-region buckets are designed for 11 nines of durability
DNo; but only because Archive objects are copied to tape in a second location
Show the answer and why
ANo; Cloud Storage is designed for at least 11 nines of annual durability whatever the storage class and location type
Correct
Cloud Storage is designed for at least 99.999999999% annual durability regardless of storage class and location type; classes and locations differ in availability and price.
BYes; colder classes are designed for lower durability in exchange for their lower price
Incorrect
High durability is a feature of all storage classes; colder classes trade availability and access costs, not durability.
CYes; only multi-region buckets are designed for 11 nines of durability
Incorrect
The 11 nines design target applies to every location type, not just multi-regions.
DNo; but only because Archive objects are copied to tape in a second location
Incorrect
Cloud Storage has no tape copies or offline retrieval for any class; all classes are served with low latency.
Durability (not losing data) and availability (being able to reach it) are different. Cloud Storage targets 11 nines of annual durability for all classes and location types, while availability and price vary by class and location.
Before a risky schema change, a DBA takes an on-demand backup of a Cloud SQL instance that uses standard backups. How long does Cloud SQL keep this on-demand backup?
AFor as long as the instance's automated backup retention setting
BFor one year, then it is deleted automatically
CUntil the next automated backup replaces it
DIndefinitely, until someone deletes the backup
Show the answer and why
AFor as long as the instance's automated backup retention setting
Incorrect
The retention setting governs automated backups; standard on-demand backups are not removed on that schedule.
BFor one year, then it is deleted automatically
Incorrect
Standard on-demand backups have no fixed retention period; Cloud SQL does not delete them after a set time.
CUntil the next automated backup replaces it
Incorrect
On-demand backups are independent of the automated schedule and are not replaced by it.
DIndefinitely, until someone deletes the backup
Correct
For standard backups, on-demand backups are stored indefinitely until they are manually deleted (or the project is deleted).
Cloud SQL offers on-demand and automated backups. Automated backup retention is set on the instance, while standard on-demand backups stay until they are deleted, which makes them useful before risky operations.
A company needs copies of a bucket's data in two regions whose pairing is not offered as a dual-region, and each copy needs its own storage class and access settings. What does Cloud Storage suggest for this?
ASeparate buckets kept in sync with cross-bucket replication
BA multi-region bucket that covers both of the regions it needs
CA single regional bucket with Object Versioning
DA regional bucket with soft delete set to 90 days
Show the answer and why
ASeparate buckets kept in sync with cross-bucket replication
Correct
Cross-bucket replication copies data between independent buckets, each with its own location, encryption, access and storage class, for pairings that dual-regions do not offer.
BA multi-region bucket that covers both of the regions it needs
Incorrect
A multi-region spreads data over a large area that the company does not choose region by region, and it is one bucket with one set of settings.
CA single regional bucket with Object Versioning
Incorrect
Versioning keeps older versions of objects in the same bucket and region; it creates no copy in a second region.
DA regional bucket with soft delete set to 90 days
Incorrect
Soft delete keeps deleted objects recoverable in the same bucket; it does not replicate data to another region.
Dual-regions and multi-regions replicate within one bucket. When the needed region pair is not available as a dual-region, or the copies need different settings, Google suggests separate buckets synchronized with cross-bucket replication (or Storage Transfer Service event-driven transfers).
A team asks what it must configure so that a BigQuery dataset in a single region keeps its data if one zone in that region is lost. What is the answer?
AA table snapshot of every table, refreshed each night
BA second dataset in another zone of the region, kept in sync by scheduled queries
CNothing; BigQuery stores data redundantly in two zones of the location
DA Cloud Storage export of each table to a multi-region bucket
Show the answer and why
AA table snapshot of every table, refreshed each night
Incorrect
Snapshots preserve table contents at a point in time; they are not needed for zonal redundancy, which BigQuery provides itself.
BA second dataset in another zone of the region, kept in sync by scheduled queries
Incorrect
BigQuery abstracts zones away; datasets are created in a region or multi-region, not a zone.
CNothing; BigQuery stores data redundantly in two zones of the location
Correct
BigQuery synchronously writes data to two zones within the location and routes queries to another zone if one fails, so a zonal outage causes no data loss.
DA Cloud Storage export of each table to a multi-region bucket
Incorrect
Exports can serve other recovery needs, but zonal redundancy is already built in.
BigQuery is a regional service that handles machine and zonal failures itself, keeping data in two zones and moving compute when a zone is disrupted. Cross-region disaster recovery and recovery from deleted or corrupted data are separate topics.
A Cloud SQL instance with high availability failed over during a zonal outage. A day later the original zone is healthy again. What is the state of the instance?
AThe instance that received the failover stays the primary, even though the original zone is back
BCloud SQL automatically fails back to the original zone as soon as it is healthy
CThe instance loses its HA configuration until it is re-created
DThe standby keeps serving only read queries until an administrator acts
Show the answer and why
AThe instance that received the failover stays the primary, even though the original zone is back
Correct
After a failover, the instance that received it continues as the primary, even after the original instance comes back online.
BCloud SQL automatically fails back to the original zone as soon as it is healthy
Incorrect
Cloud SQL does not fail back on its own when the original zone recovers; the new primary keeps serving.
CThe instance loses its HA configuration until it is re-created
Incorrect
The instance remains highly available; the original primary becomes the standby once its zone is available again.
DThe standby keeps serving only read queries until an administrator acts
Incorrect
After failover the former standby is a full primary for reads and writes; in the current HA configuration, a standby never serves reads.
Cloud SQL HA keeps a primary and a standby in two zones. On failover the standby becomes the primary and keeps that role after the original zone recovers; the old primary then becomes the standby.
A Cloud SQL instance with high availability has its primary in zone A and its standby in zone B. Reporting traffic goes to read replicas, and the team wants reads to keep working if zone A or zone B goes down. Where does Google suggest placing some of the read replicas?
AIn zone A, next to the primary
BIn a different zone, such as zone C
CIn zone B, on the standby itself
DNowhere; HA makes read replicas unnecessary
Show the answer and why
AIn zone A, next to the primary
Incorrect
During a zonal outage, traffic stops to read replicas in that zone, so replicas next to the primary go down with it.
BIn a different zone, such as zone C
Correct
Google suggests putting some read replicas in a zone other than those of the primary and standby, because traffic to replicas in a zone stops during that zone's outage.
CIn zone B, on the standby itself
Incorrect
The standby cannot serve read queries, and replicas in zone B would stop during an outage of zone B.
DNowhere; HA makes read replicas unnecessary
Incorrect
HA protects the primary's availability; it does not provide read capacity, because the standby does not serve reads.
HA and read replicas complement each other. During a zonal outage, replicas in that zone stop serving, and replicas elsewhere reconnect to the new primary, so spreading replicas across zones keeps read capacity available.