Types of backup: full, differential and incremental

The correct answer: A. Sunday's full backup and Wednesday's differential backup

✅ Why this answer

A differential backup saves everything that has changed since the last full backup. So Wednesday’s backup contains the changes from Monday, Tuesday and Wednesday together. Two steps are therefore enough: restore the full backup (Sunday), then the latest differential backup (Wednesday).

❌ Why the other options are wrong

• (B): this is the logic of incremental backup, where each backup saves only that day’s changes, so you need all the backups in order. The latest differential already contains the ones before it.
• (C): a differential backup alone contains only the changes, and does not work without the full backup it was built on.
• (D): returns the database to Sunday night, losing three days of data.

💡 Key concept

• Full: all the data. The easiest to restore, but slower and larger to create.
• Differential: the changes since the last full backup. It grows day by day, and restoring needs two backups.
• Incremental: the changes since the last backup of any type. The smallest and fastest to create, and restoring needs the whole chain.

The choice of plan depends on the maximum acceptable data loss (RPO) and the required recovery time (RTO), with restores actually tested on a regular basis.

🔗 Related facts and questions

• A quick comparison: for restoring: a full backup needs one backup, a differential needs two, and an incremental needs the full backup and everything after it, in order. For the size and time of the nightly backup the order is reversed: incremental is the smallest, then differential, then full.
• The 3-2-1 rule: three copies of the data, on two different media, with one of them off-site. With ransomware attacks, an immutable copy, or one isolated from the network, is added so the attacker cannot encrypt it along with the original.
• Backup versus replication: replication sends every change to another site at once, so it serves continuity of service. But it also sends the error and the deletion, so it does not replace a backup from an earlier point in time.
  Practice question: a table was deleted by mistake, and the database is replicated in real time to an alternate site. Is the replica enough to bring it back? → No; the deletion reached it too, and a backup from before the deletion is needed.
• RPO in numbers: one backup a night means the most data that can be lost is up to 24 hours. Backing up the transaction logs every few minutes brings that down to minutes, and allows point-in-time recovery.
  Practice question: a laboratory system cannot lose more than 15 minutes of results. Is the plan in this question enough? → No; with one backup a night it can lose up to a full day, far more than the 15-minute RPO.
• Backup versus archiving: the first is for restoring after a failure; the second keeps old, rarely used data for a long time to meet legal retention requirements.
• The priority for strengthening a plan that relies on a daily backup tape: a related question in the full bank.
• The recovery time objective: a related question in the full bank.
• The written technical procedure for restoring at night: a related question in the full bank.
• Practice question: if the nightly backups were incremental, what would be needed to restore? → Sunday’s full backup, then the Monday, Tuesday and Wednesday backups in order.