OCI Load Balancing

Load Balancer tasks:

• Service Discovery

• Health Checks

• Load Balancing

Benefits:

• Fault Tolerance and HA

• Scale

• Name abstraction: Backends don't need public IP Addresses

Supported protocols: TCP, HTTP/1.0, HTTP/1.1, HTTP/2, WebSocket

Supports SSL Termination, End-to-End SSL, SSL Tunneling, session persistence, content based routing.

Differentiators:

• Public and Private Load Balancers

• Provisioned Bandwidth: 100 Mbps, 400 Mbps, 8 Gbps

• Single Load Balancer for Layer 4 (TCP) and Layer 7 (HTTP) traffic

Public vs Private Load Balancers

Terraform Links: https://www.terraform.io/docs/providers/oci/r/load_balancer_load_balancer.html

Public Load Balancer

• Accepts traffic from the internet by using a public IP address

• It is a regional service

  • It requires either :

    • a regional subnet (recommended)

    • 2x AD-specific subnets, each in a separate AD

  • A Primary LB and a Secondary LB are created, each in a separate AD - but they are transparent to the user and you can't designate the Primary LB

  • Floating Public IP is attached to the primary Load Balancer. In case of outage, it will move to the secondary Load Balancer

Private Load Balancer

• Depending on the subnet it runs it, it can be

  • Regional

  • AD specific

• Primary and Standby each acquire a private IP

Load Balancer Components

Listener

Listener = entity that listens for incoming traffic. A load balancer supports up to 16 listeners (ports). Each listener is connected to a different Backend Set object

Terraform links: https://www.terraform.io/docs/providers/oci/r/load_balancer_listener.html

Backend Set

Backend Set = logical entity defined by a list of Backend Servers, a load Balancing Policy and a health check policy

Terraform links: https://www.terraform.io/docs/providers/oci/r/load_balancer_backend_set.html

Backend Servers

Backend servers = application server responsible to reply to incoming requets.

Terraform links: https://www.terraform.io/docs/providers/oci/r/load_balancer_backend.html

Load Balancing Policy

Policies:

• round robin (default). Each connection is distributed sequentially to each server in the backend set.

• IP Hash - uses the incoming source IP as a hash key to route non-sticky traffic to the same backend server

• least connection - routes incoming non-sticky traffic to the backend server with the fewest active connections.

How policies are applied:

• A TCP load balancer considers policy and weight criteria

• A HTTP load balancer with cookie-based session persistence forwards requests using cookie's session info

• For non-sticky HTTP request, load balancer applies policy and weight criteria.

Health Checks

Health Checks = a test to confirm availability of backed servers:

• Health Checks are activated for Backends, Backend Set, Overall Balancer.

• Types: TCP, HTTP

• Health Statuses: OK, Warning, Critical, Unknown.

• status is updated every 3 mintues