-
Load Balancing Algorithms define how incoming network requests are distributed among multiple backend servers to optimize performance, minimize latency, and ensure efficient resource usage.
-
The choice of algorithm directly impacts scalability, fairness, and reliability of the system.
- Distribution is based on predefined rules.
- Does not consider current server load or real-time conditions.
| Algorithm | Description |
|---|---|
| Round Robin | Requests are distributed sequentially to each server in a circular order. Simple and fair if servers have equal capacity. |
| Weighted Round Robin | Servers are assigned weights based on capacity (CPU, RAM). Servers with higher weights handle more requests. |
| IP Hashing | The client’s IP address is hashed to always direct the same client to the same server — useful for session persistence. |
- Adapts in real-time based on server performance, active connections, or response time.
| Algorithm | Description |
|---|---|
| Least Connections | Routes new requests to the server with the fewest active connections. Ideal when requests vary in processing time. |
| Least Response Time | Directs requests to the server with the lowest average response time and fewest connections. |
| Resource-Based (Adaptive) | Monitors server CPU, memory, and bandwidth to distribute load intelligently. Requires continuous monitoring. |
- Combines both static and dynamic methods for optimized decision-making.
| Algorithm | Description |
|---|---|
| Consistent Hashing | Maps requests (like user ID or session ID) to specific servers. Minimizes cache misses when adding/removing servers. Common in CDNs and caches. |
| Geographic Load Balancing | Routes users to the nearest data center or region to minimize latency and improve performance. |
| Random with Weighting | Randomly assigns requests while respecting server weights — balances simplicity and fairness. |
| Algorithm Type | Examples | Key Benefit | Limitation |
|---|---|---|---|
| Static | Round Robin, Weighted RR, IP Hashing | Simple and predictable | Ignores real-time server load |
| Dynamic | Least Connections, Least Response Time | Adjusts to traffic and performance | More complex, requires monitoring |
| Hybrid | Consistent Hashing, Geo Load Balancing | Balances flexibility and efficiency | Setup and tuning are complex |
| Scenario | Recommended Algorithm |
|---|---|
| Uniform traffic, equal servers | Round Robin / Weighted RR |
| Varying request sizes or durations | Least Connections / Least Response Time |
| Session persistence needed | IP Hashing / Consistent Hashing |
| Global users | Geographic Load Balancing |
| Resource-sensitive systems | Resource-Based Adaptive |