How Do Different Load‑Balancing Algorithms Impact Server Performance?

Round Robin

This method distributes incoming requests cyclically across all servers in the pool, assuming they have similar capacity; it works well when servers are homogeneous but can overload weaker servers in heterogeneous environments.

Weighted Round Robin

This algorithm adds a weight to each server based on its processing capability, allowing stronger servers to receive more requests proportionally.

Least Connection

This method assigns each new request to the server with the fewest active connections, helping avoid overload when request durations vary.

Least Connection Slow Start Time

When a server newly joins, a slow‑start period limits its connection count, gradually increasing it to prevent overload.

Weighted Least Connection

Combines server weights with current connection counts, offering a balanced distribution for heterogeneous resources.

Agent Based Adaptive Balancing

The load balancer periodically collects a load metric (0‑99) from each server and adjusts weights dynamically; if load is low it may fall back to weighted round robin, otherwise it uses adaptive weighting.

Fixed Weighted

Servers are assigned static priority weights; the highest‑weight server handles traffic unless it becomes unavailable, then the next priority takes over.

Weighted Response

Weights are calculated from health‑check response times, recomputed roughly every 15 seconds, directing traffic toward faster‑responding servers.

Source IP Hash

Requests are mapped to servers based on a hash of the client’s IP address, ensuring the same client consistently reaches the same server, though it may cause imbalance.