What Is a Load Balancer? The Tech That Keeps Sites From Crashing

Every year, lakhs of students refresh the same results page at the same second, and the server simply gives up. Every Tatkal window, thousands of people hit "book now" within the same ten seconds and the booking engine chokes. The reason some websites survive that stampede and others don't usually comes down to one unglamorous piece of infrastructure: the load balancer.

It doesn't get the attention that AI models or flashy apps do, but almost every site you use daily — your bank's net banking portal, a food delivery app, an e-commerce checkout — is sitting behind one. Here's what it actually does and why it matters more in India than most people realize.

What a Load Balancer Actually Does

A website that gets serious traffic rarely runs on just one server (a server here just means a computer dedicated to handling requests from users, usually sitting in a data center rather than on a desk). Run it on a single machine and that machine becomes a single point of failure — one crash, one overloaded CPU, and the whole site goes down for everyone.

So companies run the same application on multiple servers at once, often dozens or hundreds of identical copies. A load balancer sits in front of all of them, as the first thing a user's request touches. Its job is simple to state and hard to do well: look at every incoming request and decide which server should handle it, so that no single machine gets overwhelmed while others sit idle.

Think of it like a queue manager at a bank with ten counters open. Without one, everyone would crowd counter one while the other nine sit empty. The load balancer is the person pointing each new customer to the counter that's free.

How It Decides Who Handles Your Request

There isn't one single method — different load balancers use different logic depending on what the application needs:

• Round robin — requests are handed out to servers in simple rotation, one after another, like dealing cards.
• Least connections — the request goes to whichever server is currently handling the fewest active requests, which works better when some requests take much longer than others.
• IP hash — a user's requests are consistently routed to the same server based on their IP address, useful when a session needs to "remember" something about that user.
• Weighted distribution — more powerful servers in the group get a bigger share of traffic than weaker ones.

On top of routing traffic, a load balancer constantly checks whether each server is actually healthy. If one server starts failing or responding slowly, the load balancer quietly stops sending it new requests until it recovers — usually without the user ever noticing anything went wrong.

Why This Matters More in India Than You'd Think

India produces some of the most extreme, predictable traffic spikes anywhere in the world, and most of the famous "website crashed" moments Indian internet users complain about are really load-balancing and capacity failures. IRCTC's Tatkal booking window is the textbook example: a flood of near-simultaneous requests at exactly 10 or 11 AM, year after year, that still occasionally brings the system to its knees. Board exam result days for CBSE and state boards see similar pile-ups, as do government recruitment portals the moment an admit card or result is released, and e-commerce platforms during Big Billion Days or Flipkart's festive sales.

A well-designed load balancing setup is the difference between a site that gracefully absorbs ten times its normal traffic and one that falls over entirely. For Indian developers and startups building anything that could see a sudden spike — a ticketing platform, a UPI-linked payment flow, an exam portal — this isn't an academic networking topic, it's the thing standing between a smooth launch day and a trending-for-the-wrong-reasons moment on social media. Cloud providers with India-based data center regions, including AWS's Mumbai and Hyderabad regions and Google Cloud's Mumbai and Delhi regions, offer managed load balancing specifically so smaller Indian teams don't have to build this from scratch.

A load balancer doesn't make a server faster. It makes sure no single server has to be a hero.

Load Balancer vs CDN vs Firewall: Don't Mix Them Up

These three terms get thrown around together because they all sit "in front of" an application, but they solve different problems. A CDN caches content closer to the user's physical location to cut down loading time; a load balancer spreads live requests across multiple servers so none of them get overwhelmed; a firewall filters out malicious or unwanted traffic before it ever reaches your servers. Many real deployments use all three together, stacked in that order — firewall, then CDN, then load balancer — each handling a different part of the journey a request takes.

1. Firewall checks if the traffic is safe to let through at all.
2. CDN serves what it can from a nearby cached copy.
3. Load balancer routes whatever's left to the healthiest available server.

The Takeaway

The next time a government exam portal or a ticket-booking site buckles under load, it's rarely because "the internet is slow" — it's almost always a capacity and load-balancing problem that was underestimated at design time. For anyone building a product that could see a sudden rush of users, whether that's a startup's first product launch or a college fest's registration page, planning for your busiest possible minute, not your average one, is what separates sites that survive the spike from ones that become the day's screenshot on social media.