# Why Databases Instead of Files — Database Fundamentals

Source: https://www.skillbyai.com/en/database-fundamentals/f-why

> Explain the problems a DBMS solves compared with plain files.

## From files to a database management system

Early applications stored data in their own **files**: a CSV for customers, another for orders. As systems grew, this caused well-known problems. **Redundancy and inconsistency**: the same customer address is copied into several files and updated in only one. **Difficult access**: every new question needs a new program. **Integrity problems**: rules such as "balance must not be negative" are buried in application code. **Atomicity problems**: a crash halfway through a transfer leaves money debited but not credited. **Concurrency anomalies**: two clerks updating the same record overwrite each other. **Security**: it is hard to give users access to some data but not all. A **database management system (DBMS)** is software that stores data centrally and provides a **query language**, **integrity constraints**, **transactions** with crash recovery, **concurrency control**, **access control** and efficient **storage and indexing**. Popular relational DBMSs include PostgreSQL, MySQL, Oracle Database, SQL Server and SQLite.

## Files versus a DBMS

Separate programs with separate files on one side; many applications sharing one managed database on the other.

![Left: three program boxes each attached to its own file icon with duplicated content. Right: three program boxes all connected to a single central database cylinder with a shield.](assets/figures/database-fundamentals/section-1-map.svg) — Figure 1.1 — File-based storage compared with a shared DBMS.

## The same fact stored twice

Updating one copy and not the other leaves the data inconsistent.

```text
customers.csv
  id,name,city
  42,Asha Rao,Pune

orders.csv
  order_id,customer_id,customer_city,total
  1001,42,Pune,1499

Asha moves to Bengaluru; the program that edits customers.csv updates the city,
but orders.csv still says Pune -> two different answers to "where does Asha live?"
A DBMS stores the city once and lets every query join to it.
```

## Exam favourite

"List the disadvantages of file-processing systems" is a classic DBMS exam question. Remember them as redundancy, access difficulty, isolation of data, integrity, atomicity, concurrency and security problems.

**Quiz:** Which problem of file-based systems does a DBMS address with transactions?

- [x] Atomicity: a crash halfway through an update leaving data partially changed
- [ ] Choosing file names
- [ ] Displaying data in colour
- [ ] Compressing images

*Answer:* Atomicity: a crash halfway through an update leaving data partially changed. Transactions make multi-step updates all-or-nothing, even across crashes.