When we send a message, make an online banking transaction, watch a video, access a cloud application, use an AI platform or simply perform a Google search, the process appears almost instantaneous.
We click. The information appears.
But behind that simple interaction may be an extraordinarily complex infrastructure involving electrical power, cooling, buildings, telecommunications, networking, cybersecurity, servers, databases, storage systems and sophisticated software.
This infrastructure is the data centre.
A data centre is far more than a building filled with computers. It is a highly engineered ecosystem designed to store, process, manage, protect and deliver digital information and computing services continuously, securely and reliably.
So, what actually happens inside a data centre? Let's follow the journey from the beginning.
1. WHAT'S A DATA CENTRE FOR
At its simplest, a data centre provides the infrastructure required to operate digital services. almost everything we do online depends on data centres in one way or another. Consider some everyday activities:
- Sending an email or instant message,
- Watching Netflix or YouTube,
- Using online banking,
- Making an e-commerce purchase,
- Storing photographs in the cloud,
- Using Microsoft 365 or Google Workspace,
- Accessing an ERP system,
- Conducting a video conference,
- Using an AI application,
- Running a company's website,
- Processing financial transactions,
- Storing government or enterprise databases
Behind these services are computers and storage systems operating somewhere in the world. The data centre provides the environment in which those systems can operate.
Its five fundamental functions can be summarised as:
STORE → PROCESS → CONNECT → PROTECT → DELIVER
It stores information, processes data, connects systems and users, protects infrastructure and information, and ultimately delivers a service back to the user.
2. AN ENGINEERED ECOSYSTEM
One of the common misconceptions is that a data centre is primarily an IT facility. Well, it is not. IT equipment is certainly at the heart of the operation, but the IT equipment cannot operate without the supporting infrastructure.
A modern data centre may involve:
- Civil & Structural Engineering,
- Architectural Design,
- Electrical Engineering,
- Mechanical Engineering,
- HVAC & Cooling,
- Fire Protection,
- ICT & Telecommunications,
- Networking,
- Cybersecurity,
- Building Management Systems,,
- Data Centre Infrastructure Management
- Physical Security,
- QHSE, and
- Operations & Maintenance
This makes a data centre one of the most complex types of modern buildings.
The computers may perform the digital work, but the building and engineering systems ensure that those computers remain available.
3. IT ALL STARTS WITH ELECTRICITY
Computers need electricity.
A large data centre may require a very substantial and continuous electrical supply. Therefore, the electrical system is designed not merely to provide power, but to provide reliable, conditioned and redundant power.
Simply put :
The objective is simple: The servers should continue operating even though the normal utility supply has failed.
This is why UPS systems, batteries, generators, switchgear, transformers and distribution systems are so important.
In many facilities, redundancy is built into the electrical architecture so that maintenance or failure of one component does not necessarily interrupt the IT load.
5. NOW WE NEED TO KEEP THE COMPUTERS COOL
Electricity is not the only problem. Almost all the electrical energy consumed by computing equipment eventually becomes heat. Thousands of servers operating continuously can therefore generate enormous amounts of heat.
That heat must be removed.
A simplified cooling cycle may look like:
Traditional facilities may use air cooling, while modern high-density computing and AI facilities increasingly require more advanced cooling technologies, including direct-to-chip liquid cooling.
The principle remains the same: Remove heat from the computing equipment and reject it safely to the environment.
6. HOT AISLE AND COLD AISLE
Walk into a typical server room and you may notice something unusual. Servers are arranged so that their air intake and exhaust sides are controlled. This creates:
Cold Aisle → Server → Hot Aisle
Cold air is supplied to the front of the servers. The servers draw in the cold air. The equipment heats the air. Hot air is exhausted into the hot aisle. The cooling system then removes that heat.
This simple principle helps improve cooling efficiency and prevent hot air from mixing unnecessarily with cold supply air.
In high-density facilities, however, conventional air cooling can eventually reach practical limits.
This is one reason AI computing is driving major changes in data centre design.
7. SO, WHAT IS ACTUALLY INSIDE THE RACK?
Now we reach the part most people imagine when they hear "DATA CENTRE."
The server rack : A rack contains computing and networking equipment mounted in a standardised physical enclosure.
A typical environment may contain:
- Servers,
- Network switches,
- Storage equipment,
- Power Distribution Units,
- Network Interface Cards,
- Fibre-optic connections,
- Copper Ethernet connections, and
- Management equipment
Inside a server itself are components familiar from an ordinary computer:
CPU → RAM → Storage → Network Interface
But enterprise servers are designed for continuous operation, high performance and redundancy.
8. CPU, RAM, STORAGE AND GPU
Think of the computer as a working office.
- CPU : The Processor : The CPU performs instructions and calculations. It is effectively the general-purpose "brain" of the server,
- RAM : Working Memory : RAM temporarily holds the information that the CPU needs to access quickly. More RAM allows applications to handle larger working datasets without constantly retrieving information from slower storage,
- Storage : Storage retains information even when the computer is switched off : Modern data centres commonly use SSDs and, depending on the application, HDDs or other storage technologies,
- GPU : Graphics Processing Unit : GPUs were originally associated primarily with graphics,
Today, however, they are extremely important for:
Artificial intelligence, Machine learning, High-performance computing, Scientific modelling and Large-scale data processing
A modern AI data centre can therefore look very different from a traditional enterprise data centre because of the enormous computing and cooling requirements associated with GPU clusters.
9. HOW DO THE SERVERS COMMUNICATE?
A server by itself is not particularly useful. It needs to communicate with other servers, storage systems, users and external networks. This is where networking comes in.
A simplified architecture is:
- Network switches connect devices within the data centre,
- Routers connect different networks,
- Firewalls control and inspect traffic,
- Load balancers distribute requests across multiple servers, and
- Fibre-optic networks provide very high-speed connectivity between systems.
Once the traffic reaches the data centre, it may pass through several layers of infrastructure.
A simplified sequence is:
The firewall helps enforce security policies. The load balancer determines which available server should handle the request.
Why? Because a modern service may have hundreds or thousands of servers. Instead of sending every user to one server, requests can be distributed across multiple systems.
If one server fails, traffic can potentially be redirected to another available server.
13. WHAT DOES THE APPLICATION SERVER ACTUALLY DO?
Suppose you open an online banking application and request your account balance.
The application server receives your request. The software determines what you are asking for.
It may need to:
- Authenticate the user,
- Check permissions,
- Query a database,
- Retrieve account information,
- Apply business rules,
- Generate the requested response, and
- Return the result.
The server may therefore communicate with several other systems.
For example:
The user may see only a number on the screen.
Behind that number could be multiple computers communicating with each other.
14. DATA STORAGE
The application server is not necessarily where the permanent data resides.
Data may be stored in dedicated storage systems or distributed across multiple servers and locations.
Storage architectures can include:
- Direct-attached storage,
- Network Attached Storage,
- Storage Area Networks,
- Distributed storage,
- Object storage, and
- Cloud storage platforms.
Critical information may also be replicated.
For example:
Primary Data
↓
Replication
↓
Secondary System
↓
Backup
↓
Disaster Recovery Site
The objective is to ensure that a hardware failure, software problem or even a major facility incident does not automatically result in permanent data loss.
15. VIRTUAL MACHINES AND CONTAINERS
There is another layer that many people do not see. The physical server may run multiple virtual machines (VMs).
A hypervisor allows one physical server to host multiple virtual computing environments.
For example:
Physical Server
↓
Hypervisor
↓
VM 1 : Application
VM 2 : Database
VM 3 : Web Server
VM 4 : Management System
Modern applications may also use containers and orchestration platforms such as Kubernetes. This means that the physical hardware is only one layer of the overall computing environment. Above it sits an increasingly sophisticated software ecosystem.
16. CYBERSECURITY: SECURITY PERIMETER
A data centre must protect not only its physical equipment but also the information passing through it.
Security can involve multiple layers:
17. FIRE PROTECTION
A data centre cannot simply rely on conventional building fire protection.
The equipment is extremely valuable, and water discharge in certain environments can potentially create additional damage.
Therefore, data centres commonly use sophisticated combinations of:
- Early smoke detection,
- Fire alarm systems,
- Detection zones,
- Fire suppression systems,
- Compartmentation,
- Emergency procedures, and
- Environmental monitoring.
The actual system must be designed according to the facility, applicable standards, authority requirements and risk assessment.
18. FAILURES
This is perhaps the most important principle in data centre engineering:
Failure is expected : servers*, hard drives, power supplies, fans, network equipment, generators may fail to start, cooling equipment will require maintenance.
The objective is therefore not to create a system in which nothing ever fails. That is unrealistic but to design a system in which the failure of an individual component does not necessarily cause the service to fail.
You may hear terms such as:
N+1, 2N, 2N+1
These describe different levels of redundancy and resilience.
The appropriate architecture depends on the business requirements, availability objectives, risk profile and cost.
*18.1 Server Failures
Suppose Server A suddenly stops functioning. A monitoring system detects the failure. The load balancer may stop directing new requests to Server A.
Other servers continue processing requests. Technicians investigate the failed server.
The hardware may be repaired or replaced. Once it is tested and healthy, it can be returned to service.
To the end user, nothing may have happened.
20. MONITORING
A modern data centre is continuously monitored. Operators may monitor:
- Electrical Power,
- UPS,
- Generators,
- Temperature,
- Humidity,
- Cooling Equipment,
- Network Traffic,
- Server Health,
- Storage Capacity,
- Security Events,
- Fire Systems and
- Physical Access.
Two important systems are often encountered:
- BMS : Building Management System primarily monitors and controls building and mechanical/electrical infrastructure, and
- DCIM : Data Centre Infrastructure Management focuses more specifically on data-centre infrastructure, capacity, equipment and operational visibility.
Together with Network Operations Centres and Security Operations Centres, these systems provide operators with visibility of what is happening inside the facility.
21. DISASTER MANAGEMENT
Resilience extends beyond the individual building.
A critical application may have a Disaster Recovery (DR) site.
If the primary facility becomes unavailable, services may be transferred or recovered at another location.
This introduces concepts such as:
Backup*, Replication, Disaster Recovery, Business Continuity, Recovery Point Objective (RPO) and Recovery Time Objective (RTO)
The important distinction is that *backup is not the same thing as disaster recovery.
*A backup may preserve data. Disaster recovery concerns the ability to restore the required systems and services within an acceptable timeframe.
22. A SINGLE DATA REQUEST IS MORE LIKE A DIGITAL JOURNEY
Imagine a user in Kuala Lumpur opens an online application.
- The electrical infrastructure is supplying power,
- The UPS is conditioning critical power,
- The generators are standing by,
- The cooling system is removing heat,
- The fire detection system is monitoring the environment,
- The network is carrying traffic,
- Cybersecurity systems are analysing events,
- The BMS and DCIM are monitoring infrastructure,
- Operators are watching alarms and performance,
- Security systems are controlling physical access, and
- Maintenance teams are ensuring equipment remains operational.
- Everything must work together,
- Power, cooling and network connectivity must be available,
- Fire protection and security must function,
- Controls must communicate, and
- Servers and softwares must operate.
- Energy efficiency,
- Renewable energy integration,
- Grid infrastructure,
- Cooling technologies
- Liquid cooling
- Energy storage
- High-density electrical distribution
- Network architecture
- Automation
- AI-based infrastructure management
- Physical equipment is receiving that request,
- Electrical systems are powering the equipment,
- Cooling systems are removing the heat,
- Network systems are moving the data,
- Servers are processing instructions,
- Databases are retrieving information,
- Storage systems are preserving data,
- Cybersecurity systems are protecting the environment,
- Operators and automated systems are monitoring the entire operation, and
- Finally, the result travels back to your device.
Read More :
