A load balancer receives traffic and distributes it across a group of servers. For a website, that can mean sending requests to more than one application server instead of relying on a single machine. The goal may be to handle more traffic, maintain service during maintenance or avoid sending requests to an unhealthy server.
How requests reach a server
The balancer chooses a destination using its configured rules. A simple policy might rotate between servers; another might consider active connections or the requested hostname. Health checks help it determine which destinations should receive traffic. AWS's load-balancing overview gives a concrete example of distributing incoming traffic and monitoring target health.
For a shop, the health check should reflect whether the application can serve a useful response. A web server answering a basic request is not proof that checkout and the database are working.
Why adding a second server is not enough
Both servers need access to the data they use. Think about uploaded images, sessions, background jobs and database writes. If a visitor signs in on one server and the next request reaches another, the application must still recognise that session. Pinning a visitor to a server can help some setups, but it does not solve every failure case.
The load balancer can also become a single point of failure if the design relies on just one instance. A resilient service considers the entry point, application tier, database, storage and network together.
Scaling, failover and backups do different jobs
Scaling adds capacity. Failover moves work away from a failed component. Backups allow recovery of data. An application may need all three, and success in one does not prove the others work. Test an actual server failure and a restore rather than treating an architecture diagram as evidence.
Does your site need it?
A small website may be better served by a correctly sized managed plan. Multi-server hosting adds deployment, monitoring and data-management work. Consider it when traffic, availability requirements or maintenance windows justify that complexity.
Before a design call, record peak traffic, current CPU and memory use, the application's session and upload storage, and the amount of interruption the business can tolerate. Bring the database and background-worker requirements too. Our infrastructure service can scope those dependencies, while our VPS plans provide a starting point for individual server capacity.
Hosting that keeps up
Fast, managed UK hosting with a team that answers.
NVMe SSD, LiteSpeed and free managed migration from your current host, run by a UK support team around the clock.
