About this interactive
A database is records in tables, and on disk it is just files. Every change, a transaction, is first written to a transaction log, and the database applies the changes to its main data file as it goes.
Database backups use the log to decide what to save. A full backup copies the whole database. An incremental backup, often called a log backup, copies the transaction log since the last backup and then clears, or truncates, the part it saved. New transactions start a fresh log, so the next incremental holds only what happened since. A differential backup captures every change since the last full and does not clear the log, so each one is bigger than the last until the next full.
The lesson describes clearing the log as rolling the logs into the database, a commit. In database terms a commit is when a transaction is made permanent, and the clearing is called truncating the log. Some systems, such as Microsoft Exchange, back up and clear the logs exactly as the lesson draws it; SQL Server builds its differentials from the changed parts of the database file rather than the log. The rule to hold on to is the same everywhere: a differential is everything since the last full; an incremental is everything since the last backup.
So restoring follows the same arithmetic as files: the full plus every incremental since it, or the full plus the latest differential.
About TechKnowSurge
TechKnowSurge builds IT and cybersecurity professionals through hands-on, concept-first training built around real understanding — not memorization. Free interactive tools, structured programs, and 25+ years of real-world experience, all in one place.
Explore free tools and programs →