로드 밸런싱 및 자동 확장

Imagine your app gets super popular! Too many users trying to access a single server can slow it down or even crash it.

Load balancing is like a traffic cop for your application. It distributes incoming network traffic across multiple servers, ensuring no single server gets overwhelmed.

When a user sends a request, it first hits the load balancer. The load balancer then decides which of your available servers should handle that request.

• It acts as a single point of contact.
• It checks server health to only send traffic to working servers.
• It uses different algorithms (like round-robin) to distribute requests fairly.

Load balancers operate at different layers of the network model:

• Layer 4 (Transport Layer): Distributes traffic based on IP addresses and ports (e.g., TCP, UDP). It's fast and simple.
• Layer 7 (Application Layer): Distributes traffic based on application-level data like HTTP headers, URLs, or even cookie data. This allows for more intelligent routing decisions.

Using a load balancer brings several key advantages to your SaaS application:

• Improved Performance: Distributes load, preventing bottlenecks.
• High Availability: If one server fails, traffic is rerouted to healthy ones.
• Scalability: Easily add or remove servers without affecting users.
• Fault Tolerance: Reduces the impact of individual server failures.

What if your app has busy hours and quiet hours? Manually adding and removing servers is inefficient.

Auto-scaling automatically adjusts the number of computing resources (like servers) in your application based on demand. It ensures you have enough capacity without overspending.

Auto-scaling continuously monitors your application's metrics. When a metric crosses a set threshold, it triggers an action:

• If CPU usage is too high, add more servers.
• If network traffic drops, remove unneeded servers.

This dynamic adjustment optimizes performance and cost.

In cloud environments, auto-scaling is often managed through Auto-Scaling Groups (ASGs). An ASG defines:

• The minimum number of instances (servers) always running.
• The maximum number of instances it can scale out to.
• A desired capacity, which is the initial number of instances.

ASGs work to maintain this desired capacity and respond to scaling policies.

Auto-scaling policies define how an ASG scales. Common triggers include:

• CPU Utilization: Scale out if average CPU goes above 70%.
• Network I/O: Scale in if outbound network traffic is low.
• Custom Metrics: Based on application-specific metrics like queue length or active user count.

Policies can be simple (add N instances) or target-tracking (maintain average CPU at 60%).

These two technologies are a powerful duo! A load balancer sits in front of an Auto-Scaling Group.

• The load balancer receives all incoming traffic.
• The ASG automatically adds or removes servers based on demand.
• The load balancer automatically detects new servers added by the ASG and starts sending traffic to them.

This creates a highly available, fault-tolerant, and elastic system.

We've explored how load balancing distributes incoming traffic to prevent server overload and ensure high availability, acting as a smart traffic cop.

We also learned about auto-scaling, which dynamically adjusts your server capacity based on demand, optimizing performance and cost.

When combined, load balancers and auto-scaling groups create a robust, elastic, and highly available architecture for your SaaS application, ready to handle any traffic spike!