What It Actually Means
It also checks which servers are healthy and stops sending traffic to any that are not, which is the part that matters more than the distribution.
What It Actually Buys
Resilience, mostly. With one server, that machine failing means the system is down. With two behind a balancer, one failing means the other carries the traffic and most visitors notice nothing. For a business where downtime has a real cost, this is the main argument and it is a good one.
Maintenance without downtime. Update one server while the other serves traffic, then swap. This is how routine work stops requiring a maintenance window.
Capacity beyond one machine, which is the reason usually given and, for most businesses, the least relevant one. Very few systems genuinely exceed what a single well-configured server handles.
So the honest framing is that a load balancer is usually bought for uptime rather than for speed, and it is worth being clear about which you are paying for.
What It Requires
This is the part underestimated, because adding a second server changes assumptions the application may be making.
Sessions have to be shared. If login state is held on one server, a visitor bounced to the other appears logged out. Sessions need to live somewhere both can reach.
Uploaded files have to be shared. A file uploaded to one server does not exist on the other. Storage has to be shared or external.
Scheduled jobs must not run twice. A nightly task on both servers runs twice, which for anything sending email or charging cards is a real problem.
Deployment gets more involved. Both machines need updating, in a controlled order.
None of that is difficult. All of it has to be done deliberately, and an application built assuming one server will misbehave in ways that are hard to diagnose because they only affect some requests.
What To Ask
- What are we buying, uptime or capacity? The answer should be honest, and it is usually uptime.
- What does an hour of downtime cost us? This sets whether the arrangement is worth it. For many businesses it is not.
- Is the application ready for more than one server? Sessions, uploads and scheduled jobs specifically.
- What happens if the load balancer itself fails? It becomes the single point of failure it was meant to remove, unless the provider handles that.
- Is there a simpler option? Reliable hosting with tested backups and a fast restore covers a great deal, at considerably lower cost and complexity.
The Proportionate Position
For most small and mid-sized business systems, a single well-maintained server with monitoring, tested backups and a rehearsed recovery route delivers availability in the range the business actually needs.
A load balancer earns its place when downtime is genuinely expensive per hour, or when maintenance windows have become unacceptable. Buying one to solve a scaling problem that does not exist adds cost and a set of new requirements, and it is a common way to make a simple system complicated without making it better.
More terms are in the glossary.
