Elastic Load Balancer Deep Dive Concepts

Elastic Load Balancer Deep Dive Concepts

In Order to understand the AWS Elastic Load Balancing service, we need to understand the basic Concepts behind this. 

Listeners

Each load balancer has one or more Listeners, that receives the request from client. Each Listener is configured with a protocol, a port for front end connection (Client to Load balancer) and a port for backend connection (Load Balancer to EC2 instance).

Application and Classic Load Balancer both supports following protocols:

• HTTP
• HTTPS
• TCP
• SSL

Listener Rule:

Listener Rule is the set of rule that you define for your Application Load Balancer, this rule determines, how the load balancer routes requests to the target in one or more target groups.

A Listener rule are made up of following:

A Rule Priority: Rules are read in priority order, from lower priority to Higher value.

Action Rule: Each action rule has a type and its associated target group. Currently only supported type is forward, which forward requests to target group.

Rule Condition: It is the set of condition, that has to be met, if the rule condition is met, then action is taken.  A patch Condition rule can also be defining, which forwards the request to different target group based on URL, also called as path based Routing.   

Targets

Targets are the registered EC2 endpoints that finally receive the traffic and server client requests.

Below is some method, by which EC2 instances can be registered in targets.

• An Instance, targets are specified by instance ID
• An IP of the EC2 instances going to be registered in targets
• Subnets of the VPC for target group.
• 0.0.0/8 (RFC 1918)
• 64.0.0/10 (RFC 6598)
• 16.0.0/12 (RFC 1918)
• 168.0.0/16 (RFC 1918)

Target Groups

A number of Similar EC2 instance are grouped together, with the help of targets Groups. These target Groups are used in Listener rule. Health check setting are also defined per target groups basis. Once this is defined, LB do the health check for targets in targets groups and route the traffic to targets in particular target groups which are healthy.

Elastic load balancer configuration

In this section we will learn about the idle connection timeout, Cross-zone load balancer, Connection draining, Proxy Protocol, Sticky session and health checks.

Idle Connection timeout:

Once any request is sent by client to Load balancer, it maintains a front end connection (between client to Load balancer) and from Load balancer to Backend server. For each connection, load balancer manages an idle connection time out that is triggered when no data is sent over the connection for specified period. Once Idle time out is expired and no data has been sent tor received, load balancer will close the connection.

By default, Elastic Load Balancer sets idle time out to 60 secs for both connections. This values can be changed based on traffic type or lengthy operation.