AWS::GlobalAccelerator Construct Library¶---
AWS Global Accelerator (AGA) is a service that improves the availability and performance of your applications with local or global users.
It intercepts your user’s network connection at an edge location close to them, and routes it to one of potentially multiple, redundant backends across the more reliable and less congested AWS global network.
AGA can be used to route traffic to Application Load Balancers, Network Load Balancers, EC2 Instances and Elastic IP Addresses.
For more information, see the AWS Global Accelerator Developer Guide.
Here’s an example that sets up a Global Accelerator for two Application Load Balancers in two different AWS Regions:
# Create an Accelerator accelerator = globalaccelerator.Accelerator(self, "Accelerator") # Create a Listener listener = accelerator.add_listener("Listener", port_ranges=[globalaccelerator.PortRange(from_port=80), globalaccelerator.PortRange(from_port=443) ] ) # Import the Load Balancers nlb1 = elbv2.NetworkLoadBalancer.from_network_load_balancer_attributes(self, "NLB1", load_balancer_arn="arn:aws:elasticloadbalancing:us-west-2:111111111111:loadbalancer/app/my-load-balancer1/e16bef66805b" ) nlb2 = elbv2.NetworkLoadBalancer.from_network_load_balancer_attributes(self, "NLB2", load_balancer_arn="arn:aws:elasticloadbalancing:ap-south-1:111111111111:loadbalancer/app/my-load-balancer2/5513dc2ea8a1" ) # Add one EndpointGroup for each Region we are targeting listener.add_endpoint_group("Group1", endpoints=[ga_endpoints.NetworkLoadBalancerEndpoint(nlb1)] ) listener.add_endpoint_group("Group2", # Imported load balancers automatically calculate their Region from the ARN. # If you are load balancing to other resources, you must also pass a `region` # parameter here. endpoints=[ga_endpoints.NetworkLoadBalancerEndpoint(nlb2)] )
The Accelerator construct defines a Global Accelerator resource.
An Accelerator includes one or more Listeners that accepts inbound connections on one or more ports.
Each Listener has one or more Endpoint Groups, representing multiple geographically distributed copies of your application. There is one Endpoint Group per Region, and user traffic is routed to the closest Region by default.
An Endpoint Group consists of one or more Endpoints, which is where the user traffic coming in on the Listener is ultimately sent. The Endpoint port used is the same as the traffic came in on at the Listener, unless overridden.
Types of Endpoints¶
There are 4 types of Endpoints, and they can be found in the
Application Load Balancers
Network Load Balancers
Elastic IP Addresses
Application Load Balancers¶
# alb: elbv2.ApplicationLoadBalancer # listener: globalaccelerator.Listener listener.add_endpoint_group("Group", endpoints=[ ga_endpoints.ApplicationLoadBalancerEndpoint(alb, weight=128, preserve_client_ip=True ) ] )
Network Load Balancers¶
# nlb: elbv2.NetworkLoadBalancer # listener: globalaccelerator.Listener listener.add_endpoint_group("Group", endpoints=[ ga_endpoints.NetworkLoadBalancerEndpoint(nlb, weight=128 ) ] )
# listener: globalaccelerator.Listener # instance: ec2.Instance listener.add_endpoint_group("Group", endpoints=[ ga_endpoints.InstanceEndpoint(instance, weight=128, preserve_client_ip=True ) ] )
Elastic IP Addresses¶
# listener: globalaccelerator.Listener # eip: ec2.CfnEIP listener.add_endpoint_group("Group", endpoints=[ ga_endpoints.CfnEipEndpoint(eip, weight=128 ) ] )
Client IP Address Preservation and Security Groups¶
When using the
preserveClientIp feature, AGA creates
Elastic Network Interfaces (ENIs) in your AWS account, that are
associated with a Security Group AGA creates for you. You can use the
security group created by AGA as a source group in other security groups
(such as those for EC2 instances or Elastic Load Balancers), if you want to
restrict incoming traffic to the AGA security group rules.
AGA creates a specific security group called
GlobalAccelerator for each VPC
it has an ENI in (this behavior can not be changed). CloudFormation doesn’t
support referencing the security group created by AGA, but this construct
library comes with a custom resource that enables you to reference the AGA
endpointGroup.connectionsPeer() to obtain a reference to the Security Group
which you can use in connection rules. You must pass a reference to the VPC in whose
context the security group will be looked up. Example:
# listener: globalaccelerator.Listener # Non-open ALB # alb: elbv2.ApplicationLoadBalancer # Remember that there is only one AGA security group per VPC. # vpc: ec2.Vpc endpoint_group = listener.add_endpoint_group("Group", endpoints=[ ga_endpoints.ApplicationLoadBalancerEndpoint(alb, preserve_client_ip=True ) ] ) aga_sg = endpoint_group.connections_peer("GlobalAcceleratorSG", vpc) # Allow connections from the AGA to the ALB alb.connections.allow_from(aga_sg, ec2.Port.tcp(443))