Task 3.1: Design solutions for backup and disaster recovery

• AAzure Backup with Cross Region Restore of every VM into West Europe

  Incorrect

  Restoring from daily or hourly recovery points loses far more than a few minutes, and restoring each VM takes too long for one hour.

• BAvailability zones for every VM in North Europe

  Incorrect

  Zones protect against a datacenter failure inside one region, not the loss of the whole region.

• CManaged disk snapshots copied to West Europe every night

  Incorrect

  Nightly copies lose up to a day of data and need the VMs rebuilt by hand.

• DAzure Site Recovery replication to West Europe, with a recovery plan

  Correct

  Site Recovery replicates VMs continuously, and a recovery plan fails over groups in order and can run scripts or runbooks.

Site Recovery continuously replicates Azure VMs to a secondary region. Recovery plans group machines, fail them over in a defined order and run scripts or Azure Automation runbooks, so a multi-tier app comes up in order within the RTO.

• AGeo-redundant storage for the vault

  Incorrect

  Geo-redundancy copies data to another region; it doesn't stop an authorized user from deleting it.

• BAn immutable vault, with immutability locked

  Correct

  A locked immutable vault blocks operations that would delete recovery points before expiry, such as reducing retention, and can't be turned off.

• CA CanNotDelete resource lock on the vault

  Incorrect

  A delete lock doesn't block updates, so a policy's retention can still be lowered, and it adds no second approver.

• DCustomer-managed keys for encryption of backup data

  Incorrect

  Customer-managed keys control encryption; they don't block deletion or retention changes.

• EMulti-user authorization with a Resource Guard owned by the security team

  Correct

  Critical operations protected by a Resource Guard need permission on the guard, which the backup admins don't hold.

Immutable vaults prevent operations that could lead to loss of recovery points, and locking immutability makes that irreversible. Multi-user authorization adds a Resource Guard, typically in another subscription or tenant, so critical operations need a second party's approval.

• AA long-term retention policy that keeps monthly backups for seven years

  Correct

  Long-term retention keeps weekly, monthly or yearly full backups for up to 10 years, alongside the short-term point-in-time window.

• BA point-in-time restore window of seven years

  Incorrect

  The point-in-time window is limited to 35 days at most.

• CGeo-restore from the geo-replicated backups

  Incorrect

  Geo-restore uses the latest geo-replicated backups for a regional outage; it doesn't keep seven years of monthly copies.

• DAn active geo-replica in a second region

  Incorrect

  A geo-replica mirrors current data, including mistakes, and keeps no historical copies.

Short-term backup retention supports point-in-time restore for 1 to 35 days. Long-term retention stores full backups weekly, monthly or yearly for up to 10 years in separate storage, restorable as a new database.

• AGeo-redundant storage without Cross Region Restore, changeable at any time

  Incorrect

  Without Cross Region Restore the secondary copy isn't available to you on demand, and redundancy is locked once backups exist.

• BZone-redundant storage, set before the first backup is configured

  Incorrect

  ZRS keeps all copies in the primary region, so a regional outage leaves nothing to restore elsewhere.

• CGeo-redundant storage with Cross Region Restore, set before any backup

  Correct

  Cross Region Restore lets you restore from the secondary region on demand, and the vault's redundancy can't change once backups exist.

• DLocally redundant storage, with a second vault added in the paired region later

  Incorrect

  A new vault in another region holds none of the existing recovery points.

A vault's storage replication type must be set before any backup is configured. With GRS and Cross Region Restore enabled, recovery points replicated to the secondary region can be restored there at any time, for example during a primary region outage.

• AVaulted backup for blobs in an Azure Backup vault

  Correct

  Vaulted backup copies the data to the Backup vault, outside the source account, on a schedule with its own retention.

• BOperational backup for blobs with point-in-time restore

  Incorrect

  Operational backup is local: the data stays in the source account, so losing the account loses the backups.

• CObject replication to a second account in the same region

  Incorrect

  Object replication copies changes, including encrypted overwrites, and the destination is just another account.

• DBlob snapshots taken by a daily script

  Incorrect

  Snapshots live in the same account as the base blob, so they're lost with it.

Azure Backup offers operational backup, a local solution that keeps data in the source storage account, and vaulted backup, which transfers data to a Backup vault on a schedule. Only vaulted backup protects against loss or compromise of the source account.

• AMicrosoft Azure Backup Server in the branch office

  Incorrect

  Azure Backup Server is a separate server, which the branch can't add.

• BAzure Site Recovery replication of the file servers

  Incorrect

  Site Recovery doesn't keep 90 days of file versions.

• CAzure File Sync with cloud tiering

  Incorrect

  File Sync caches and tiers files to an Azure file share; it isn't a backup with retention points.

• DThe Microsoft Azure Recovery Services (MARS) agent on each file server

  Correct

  The MARS agent runs on the Windows machine itself and backs up files and folders directly to a Recovery Services vault.

The MARS agent backs up files, folders and system state from Windows machines directly to a Recovery Services vault, with no separate backup server, on a schedule with its own retention.

• ARun a test failover into an isolated virtual network

  Correct

  A test failover creates VMs from replicated data in a network you choose while replication continues and production is untouched.

• BRun a planned failover and then fail back

  Incorrect

  A planned failover shuts down the source VMs, which stops production for the drill.

• CRestore the VMs from Azure Backup into the production network

  Incorrect

  Restoring into the production network risks IP and name conflicts and doesn't test the Site Recovery setup.

• DDisable replication, start the target VMs, then re-enable replication

  Incorrect

  Disabling replication removes protection during the drill and needs a full resync afterwards.

A disaster recovery drill in Site Recovery is a test failover: VMs are created in a test network, ideally isolated from production, without data loss or impact on replication, and are cleaned up afterwards.

• AAzure VM backup with an Enhanced policy that runs every four hours

  Incorrect

  VM-level backup takes snapshots hours apart and has no log backups, so it can't restore one database to a point within 15 minutes.

• BNative SQL Server backups written by an Agent job to a file share on the VM

  Incorrect

  Self-managed jobs and storage are what the team wants to avoid, and backups on the same VM share its fate.

• CAzure Site Recovery replication of the VM

  Incorrect

  Site Recovery is for failover of the current state, not restoring a database to an earlier point in time.

• DAzure Backup for SQL Server in Azure VMs, with log backups every 15 minutes

  Correct

  Workload-aware SQL backup takes full, differential and log backups, giving a 15-minute RPO and point-in-time restore per database.

Azure Backup's SQL Server in Azure VM solution is workload-aware: it supports full, differential and log backups, a 15-minute RPO with frequent log backups, point-in-time recovery and database-level restore, with no backup infrastructure to manage.

• APeriodic backup with a four-hour interval

  Incorrect

  Periodic backups restore only to backup points and, depending on the setup, can require a support request.

• BThe analytical store with a 30-day retention

  Incorrect

  Analytical store is for analytics on a columnar copy, not for restoring the transactional data.

• CA second write region for the account

  Incorrect

  Another region replicates the corruption too; it doesn't keep earlier states.

• DContinuous backup with the 30-day tier

  Correct

  Continuous backup lets you restore to any timestamp within the tier's retention window yourself.

Continuous backup mode records changes so that an account, database or container can be restored to any point in time within the retention of the chosen tier (7, 30 or 35 days), using the portal, CLI or PowerShell.

• AThe data is gone, because soft delete must be turned on before anything is deleted

  Incorrect

  Soft delete is enabled by default for vaults; nothing had to be turned on.

• BThe data can be recovered only through a Microsoft support request

  Incorrect

  Undeleting a soft-deleted backup item is a self-service operation in the vault.

• CIt's soft-deleted and can be undeleted; retention is 14 days by default

  Correct

  Soft delete is on by default and keeps deleted backup data for 14 days, extendable to 180, so it can still be recovered.

• DThe data is kept for 90 days, the minimum soft delete retention

  Incorrect

  The default soft delete retention is 14 days, not 90.

With secure by default, soft delete is automatically enabled for recovery points, backup items and vaults. Deleted backup data is kept for the soft delete retention period, 14 days by default and extendable up to 180 days, and can be undeleted during that time.