From File Systems to Relational

Photo by Caspar Camille Rubin / Unsplash

Databases: A Quick Overview

Before databases existed, applications each managed their own data files. Every program had its own format, its own storage, and no way to share data with anything else. This caused a lot of duplication, inconsistency, and maintenance headaches.

The database approach solved this by introducing a Database Management System (DBMS) — software that centralises data storage, manages access, and separates the data itself from the programs that use it. Instead of each app owning its data, everything goes through one managed system.

How a Database is Structured

Modern databases follow a three-level architecture:

External level — what each user or application sees (their personal view of the data)
Conceptual level — the full logical picture of all the data and how it relates
Internal level — how the data is physically stored on disk

This separation means you can change how data is stored without breaking the applications on top, and vice versa.

How Data Models Evolved

Data models went through several generations:

Hierarchical — tree structure, each child has one parent; rigid and hard to maintain
Network — allowed more complex relationships, but required navigating physical pointers
Relational — introduced by E. F. Codd in 1970; data stored in tables, queried using SQL; became the dominant model
Object-Oriented and Object-Relational — extended the relational model to handle more complex, nested data structures

Designing a Database

Database design goes through three phases: conceptual (ER diagrams, entities and relationships), logical (turning that into tables and keys), and physical (deciding actual storage structures, indexes, and hardware).

Good design also means normalisation — organising tables to remove redundancy and prevent update errors. The main normal forms go from 1NF up to BCNF and 4NF, each removing a different type of data anomaly. When performance becomes a concern, controlled denormalisation can be applied to speed up reads at the cost of some extra write complexity.

Security

Database security operates in layers: access control (who can read or write what), views (showing only what a user is allowed to see), encryption (protecting data at rest and in transit), and physical redundancy via RAID for hardware fault tolerance. SQL injection is a key application-level threat, mitigated by using parameterised queries.

The overall story is one of progressive decoupling — moving from tightly coupled file systems toward clean separation of data, logic, and storage, with security and integrity built in at every layer.

When Grief Behaves Like a Black Hole

I came across a framework that maps grief onto astrophysics — gravitational collapse, event horizons, time dilation — and my first reaction was that it sounds like academic overreach. But the more I sat with it, the more it made sense. Not because the metaphors are poetic. Because they describe something I

Malaysia's Postcolonial Problem Is Not Ancient History

There is a version of Malaysian history that gets taught in schools, repeated in official ceremonies, and printed in textbooks. In that version, independence happened on 31 August 1957, Malaysia is one nation that formed naturally, and the country has been moving forward ever since. That version leaves out quite

Creative Thinking Is Just Problem-Solving Under Pressure

I will be honest: I almost did not make it to this activity at all. This semester has been one of the harder ones. My car broke down. I have been dealing with financial strain. I missed classes and presentations. I am autistic and schizophrenic, and when life gets unstable,

Education vs. Corporate Training: What Are Schools Actually For?

There is a difference between education and training, and it matters. Education is about developing a person — their curiosity, their capacity to think, their ability to engage with the world on their own terms. Training is about making a person useful to someone else's system. The problem is