What a DBMS does

Why it matters

Every application that outlives a single run needs to keep data somewhere. Writing it to files works until two users edit at once, the machine crashes halfway through a save, someone needs a report the code was not written for, or the data no longer fits in memory. A database management system exists to solve exactly those problems once, for everyone. Knowing what it does for you, and how it is built, is the foundation for every lesson in this stage and for most of the database questions in interviews.

The idea

A database is an organised collection of data. A database management system (DBMS) is the software that stores it, answers questions about it, keeps it consistent, and protects it from concurrent users and crashes. Programs never touch the data files directly; they send requests (usually SQL) and the DBMS does the work.

What goes wrong with plain files

Problem with files	What the DBMS provides
Data duplicated across files drifts apart	One shared, integrated store with constraints
Every new question needs a new program	A declarative query language: say what you want, not how to find it
Two users editing at once overwrite each other	Concurrency control: transactions that behave as if run alone
A crash mid-save leaves half-written data	Atomicity and recovery from a log
Anyone with file access sees everything	Users, roles and permissions per table and column
Changing the file format breaks every program	Data independence: storage can change without changing queries
Searching means reading everything	Indexes and a query optimizer

Inside a DBMS

Schema, instance and data independence

The schema is the structure: tables, columns, types, constraints. It changes rarely. The instance is the data in the database at one moment, which changes constantly. Databases describe data at three levels (the three-schema architecture): the external level (views tailored to each user or application), the conceptual level (the logical structure of all data) and the internal level (how it is physically stored). This separation gives data independence: physical independence means you can add an index, change file layout or move to faster disks without changing any query; logical independence means you can add a column or split a table and keep old applications working through views. It is the single most important idea a DBMS provides.

The languages

Language	Purpose	Examples
DDL (data definition)	Define and change the schema	CREATE TABLE, ALTER TABLE, DROP INDEX
DML (data manipulation)	Read and change the data	SELECT, INSERT, UPDATE, DELETE
DCL (data control)	Grant and revoke permissions	GRANT, REVOKE
TCL (transaction control)	Group changes into units	BEGIN, COMMIT, ROLLBACK, SAVEPOINT

SQL contains all four. It is declarative: a query describes the result, and the DBMS's optimizer decides the steps, which is why the same query can get faster after someone adds an index without anyone rewriting it.

Who touches the database

End users through applications; application developers writing queries; and database administrators who manage schema, permissions, backups, performance and recovery. In most companies today the DBA role is shared between platform teams and managed cloud services, but the responsibilities have not gone away.

Trade-offs

A DBMS buys correctness under concurrency and crashes, flexible querying and independence from physical layout, at the cost of overhead for each operation, operational work (backups, upgrades, tuning) and a schema that must be designed and evolved. For a configuration file or a cache, plain files or a key-value store are simpler; for shared business data that must stay correct, a DBMS is the default.

In an interview

The classic opener "why use a DBMS instead of files?" wants the list from the comparison table in your own words: integration without redundancy, declarative queries, concurrency control, crash recovery, security and data independence. Follow with one concrete failure: "two clerks updating the same seat booking file would both sell the last seat; a DBMS serialises those transactions."