Create a SIP Overlay - Real-Time Communication on AWS

Create a SIP Overlay

AWS has a robust, scalable and redundant network backbone that provides connectivity between different Regions. When a network event, such as a fiber cut, degrades an AWS backbone link, traffic is quickly failed over to redundant paths using network level routing protocols, such as BGP. This network level traffic engineering is a black box to AWS customers and most do not even notice these failover events. However, customers that run real-time workloads, such as voice, high quality video, and low latency messaging, do sometimes notice these events. So, how can an AWS customer implement their own traffic engineering on top of what is provided by AWS at the network level? The solution is deploying SIP infrastructure at many different AWS Regions. As part of the call control features, SIP also provides the ability to route calls through specific SIP proxies.

Using SIP routing to override network routing

Figure 9: Using SIP routing to override network routing

In Figure 9, SIP infrastructure (represented by green dots) is running in all four US Regions. The blue lines represent a fictional depiction of the AWS backbone. If no SIP routing is implemented, a call originating in the US west coast and destined for the US east coast goes over the backbone link that is directly connecting the Oregon and Virginia regions. The diagram shows how a customer might override the network level routing and make the same call between Oregon and Virginia routed through California using SIP routing. This type of SIP traffic engineering can be implemented using SIP proxies and media gateways based on network metrics such as SIP retransmissions and customer specific business preferences.