Q16Cloud Computing
Question
4 marks
Discuss load balancing in Cloud Computing.
Answer
A technical exploration of Cloud Load Balancing, detailing how Layer 4 and Layer 7 routing algorithms distribute massive TCP/HTTP traffic across server clusters to mathematically prevent catastrophic bottlenecking.
If a viral web application suddenly receives 1,000,000 requests per second, a single Virtual Machine will suffer a catastrophic CPU meltdown and violently crash. Cloud architectures prevent this by deploying a Load Balancer—a highly optimized, reverse-proxy network appliance that sits in front of a massive Auto-Scaling group of servers, mathematically distributing the incoming traffic matrix.
Architectural Routing Layers
- Layer 4 (Network Load Balancing): Operates at the absolute lowest TCP/UDP level. It is blisteringly fast. It mathematically evaluates the incoming IP addresses and ports and routes traffic without ever looking at the actual data payload. Ideal for raw database traffic.
- Layer 7 (Application Load Balancing): Operates at the HTTP/HTTPS level. It is highly intelligent. It mathematically decrypts the SSL packet, reads the URL path (e.g.,
/apivs/images), and violently routes the traffic to specialized server clusters optimized for that specific task.
Mathematical Distribution Algorithms
- Round Robin: The absolute simplest algorithm. Request 1 goes to Server A, Request 2 to Server B, Request 3 to Server C. It assumes all servers have perfectly equal CPU power.
- Least Connections: The Load Balancer aggressively monitors the open TCP sockets on every server. It routes new traffic to the server currently mathematically holding the fewest active connections.
- Sticky Sessions (Session Affinity): A specialized routing protocol. Once User A connects to Server B, the Load Balancer mathematically drops a cookie in the user's browser, violently forcing all future requests from User A to return strictly to Server B, preserving in-memory session data.