Introduction
As applications grow, a single server can no longer handle all incoming traffic reliably. Load balancers solve this by distributing requests across a pool of backend instances, improving both scalability and availability. Cloud providers offer managed load balancing services so you never have to run or patch the balancing software yourself.
Cricket analogy: A single server handling all traffic is like one net bowler trying to bowl to the entire squad during nets; a load balancer is like a coach who distributes bowling duties across several net bowlers, and cloud providers hand you a ready-made rotation system instead of making you train the bowlers yourself.
Explanation
Load balancers operate at different layers of the networking stack. A Layer 4 load balancer works at the transport layer, distributing traffic based on IP address and TCP or UDP port information without inspecting the content of the request. It is fast and protocol-agnostic, making it a good fit for non-HTTP traffic or extremely high-throughput scenarios. A Layer 7 load balancer works at the application layer, understanding protocols like HTTP and HTTPS. Because it can read the actual request, a Layer 7 load balancer can route based on URL path, hostname, headers, or cookies, enabling patterns like routing /api requests to one service and /images requests to another.
Cricket analogy: A Layer 4 load balancer is like an usher directing spectators to gates purely by ticket number without reading the ticket's seat details, fast but basic; a Layer 7 load balancer is like an usher who reads the actual seat and section printed on the ticket to route VIP box holders differently from general stand fans.
Regardless of layer, a load balancer must know which backend instances are healthy. It does this through health checks: periodic probes, such as an HTTP GET to a /health endpoint or a TCP connection attempt, sent to each registered instance. If an instance fails enough consecutive checks, the load balancer marks it unhealthy and stops routing traffic to it until it recovers, preventing users from being sent to a broken server.
Cricket analogy: Health checks are like a coach periodically checking each bowler's fitness with a quick fitness test before every session; if a bowler fails enough consecutive tests, the coach rests them from the bowling rotation until they're cleared fit again, protecting the team from a breakdown mid-over.
Example
Layer 7 (HTTP) load balancer routing rules:
Client request
|
[Load Balancer]
|-- path /api/* -> Target Group: api-servers (health check: GET /api/health)
|-- path /images/* -> Target Group: image-servers (health check: GET /images/ping)
|-- path /* -> Target Group: web-servers (health check: GET /)
Health check result per instance:
web-server-1: healthy (200 OK, last 3 checks)
web-server-2: unhealthy (timeout, last 3 checks) -> removed from rotation
Analysis
In this example, the Layer 7 load balancer inspects the URL path of each request to decide which target group should handle it, something a Layer 4 balancer could not do since it cannot see HTTP-level data. Meanwhile, health checks run independently of client traffic: web-server-2 is automatically pulled out of rotation after failing three consecutive checks, so real users never get routed to it. This combination of content-aware routing and continuous health monitoring is what lets cloud applications stay available even as individual instances fail or are replaced.
Cricket analogy: The scoring app's router reads whether a request is for live-score updates or replay highlights and sends each to the right service, something a basic ticket-scanner-style system couldn't do; meanwhile a struggling commentary server is quietly pulled from rotation after failing checks, so fans watching live never notice.
Key Takeaways
- Layer 4 load balancers route based on IP and port, without inspecting request content.
- Layer 7 load balancers understand HTTP/HTTPS and can route on path, host, or headers.
- Health checks continuously probe backend instances and remove unhealthy ones from rotation.
- Load balancing improves both scalability (spreading load) and availability (avoiding failed instances).
Practice what you learned
1. What distinguishes a Layer 7 load balancer from a Layer 4 load balancer?
2. What is the purpose of a health check in a load balancer?
3. Which layer does a Layer 4 load balancer operate at?
4. What happens when an instance fails enough consecutive health checks?
Was this page helpful?
You May Also Like
Auto-Scaling Basics
Learn the difference between horizontal and vertical auto-scaling and how scaling policies use metrics like CPU utilization to trigger scaling actions.
High-Availability Design
Design systems that stay up through redundancy and automatic failover, and understand what each additional 'nine' of uptime really costs.
Cloud Networking Basics
Learn the core building blocks that connect and isolate resources in the cloud: VPCs, subnets, route tables, and gateways.
Content Delivery Networks (CDNs)
Learn how CDNs cache content at distributed edge locations to cut latency, and how cache hits and misses work.