What is Load Balancer?
Turkish: Load Balancer
A load balancer distributes incoming network traffic across multiple servers to ensure high availability and performance.
What is a Load Balancer?
A load balancer distributes client traffic across multiple application servers, containers, or service instances. It prevents one server from becoming the bottleneck, routes traffic away from unhealthy nodes, and supports maintenance without taking the whole application offline.
Load balancing can work at layer 4 for TCP/UDP traffic or at layer 7 for HTTP requests. Common algorithms include round robin, least connections, weighted routing, and IP hash. Health checks remove failing instances from the pool; in some architectures, TLS termination, sticky sessions, path-based routing, and rate limiting also happen at the load balancer.
Design Considerations
- Store session state in a shared store rather than local server memory.
- Align timeout, keep-alive, and connection limits with backend behavior.
- Make health check endpoints reflect real application readiness.
- Monitor traffic distribution with logs, metrics, and traces.
A load balancer is a core part of horizontal scaling, but a bad configuration can spread failure to all traffic. CDN covers static and edge delivery, microservice explains service-level architecture, and observability helps verify behavior in production.
Related Terms
A CDN caches static content on edge servers near users, reducing latency, bandwidth pressure, and load on the origin server.
MicroserviceMicroservice architecture designs a large application as small services that can be developed, deployed, and scaled independently.
NginxNginx is an event-driven web server that serves static content, proxies requests, and distributes traffic across backend servers.
Reverse ProxyA reverse proxy sits in front of backend servers, routing client traffic while handling TLS termination, caching, security, and load balancing.
ThroughputThroughput measures the number of requests, transactions, or data units a system can process per second, minute, or hour.