AWS Global Accelerator

Your network latency is driven by the number of networks your user data needs to hop and the bandwidth available along the path to your AWS application origins. These network variables create opportunities for internet congestion to delay connections and lose data. AWS Global Accelerator combines advanced networking features with the dedicated AWS Global Network to improve your application network performance by up to 60%. TCP connections are terminated at the AWS Edge location closest to your users, instead of at your endpoint, accelerating data transfers globally. Once on the AWS network, automated routing directs your user traffic to the most performant AWS origins in Regions and/or Availability Zones. For UDP workloads, the AWS network provides the global capacity needed to avoid packet loss and jitter during traffic spikes.

You need to build your architecture with resiliency and availability in mind. This can mean running your application in a single AWS Region across multiple Availability Zones or across multiple AWS Regions. Wherever you route your traffic on the AWS network, with Global Accelerator, failover between application origins happens automatically and within seconds. If Global Accelerator detects failure of your application endpoint it instantly triggers traffic re-routing to the next available, closest endpoint in another AZ or AWS Region. Your users are redirected without needing to anylist new IP addresses or update their DNS cache. 

As your application grows, the number of origins and IP addresses that you need to manage increases and becomes burdensome. Your updates and changes refresh the DNS cache for your application, risking lower availability due to firewalls, hardcoded devices and allow-list requirements. AWS Global Accelerator simplifies global traffic management by providing 2 static anycast IP addresses that only need to be allow-listed by users once. Behind these IP address you can add or remove AWS origins, opening up uses such as endpoint failover, scaling, or testing without any user-side changes. For A/B testing or blue green deployment, use traffic dials or endpoint weights to customize how much traffic is going to each endpoint. You can bring your own IP addresses (BYOIP) to AWS Global Accelerator or use static IP addresses from the Amazon IP address pool.

Exposing your AWS origins, such as Application Load Balancers or EC2 instances, to public internet traffic creates an opportunity for malicious attacks. AWS Global accelerator decreases the risk of attack by masking your origin behind two static entry points. These entry points are protected by default from Distributed Denial of Service (DDoS) attacks with AWS Shield. AWS Global Accelerator creates a peering connection with your Amazon Virtual Private Cloud using private IP addresses, keeping connections to your internal Application Load Balancer or private EC2 instance off the public internet.