How client affinity works in Global Accelerator
If you have stateful applications that you use with a standard accelerator, you can configure client affinity to have Global Accelerator direct all requests from a user at a specific source (client) IP address to the same endpoint resource. Choosing this option maintains client affinity for your users.
By default, client affinity for a standard listener is set to None and Global Accelerator distributes traffic equally between the endpoints in the endpoint groups for the listener.
Global Accelerator uses a consistent-flow hashing algorithm to choose the optimal endpoint for a user's connection. If you configure client affinity for your Global Accelerator resource to be None, Global Accelerator uses the 5-tuple properties—source IP, source port, destination IP, destination port, and protocol—to select the hash value. Next, it chooses the endpoint that provides the best performance. If a given client uses different ports to connect to Global Accelerator and you've specified this setting, Global Accelerator can't ensure that connections from the client are always routed to the same endpoint.
If you want to maintain client affinity by routing a specific user—identified by their source IP address—to the same endpoint each time they connect, set client affinity to Source IP. When you specify this option, Global Accelerator uses the 2-tuple properties—source IP and destination IP—to select the hash value and route the user to the same endpoint whenever they connect. Additionally, Global Accelerator honors client affinity by routing all connections with the same source IP address to the same endpoint group.
On occasion, network maintenance or disruptions created by variations in internet traffic routing can cause client traffic to shift to different Global Accelerator edge locations. When this happens, if the edge location that now serves the client traffic prefers a different AWS Region, then client affinity is not guaranteed to be maintained.
In addition, be aware that when you've set endpoint weights in your accelerator, in specific, limited scenarios, Global Accelerator overrides those weights, to help ensure availability. When Global Accelerator is load balancing traffic across endpoints in an endpoint group, it must, in certain circumstances, choose between preserving availability for client traffic and abiding by endpoint weights. For example, with accelerators where the client IP address is preserved, Global Accelerator might need to override an endpoint weight setting to help avoid connection collisions.
