Is Windows Azure Traffic Manager a single point of failure?

Question

There is not much documentation around for the Windows Azure Traffic Manager.

Its principal use (for me) is to provide resilience for a service by aggregating multiple service instances behind a single Traffic Manager policy.

Can anyone explain whether the Azure Traffic Manager represents a single point of failure itself? How likely is it that the Azure Traffic Manager can fail? Does it mitigate the risk of failure?

Answer 1

Based on the information published here, Windows Azure Traffic Manager (WATM) is still in CTP, so the information below may change when it is released as final. As most of Windows Azure Services have 99.9x% availability SLA so this may also have same/similar SLA.

to make WATM highly available, there are several major components i.e. DNS resolution, Service health monitoring, management etc added, these components are deployed to multiple data centers worldwide and highly available. This design solves problems of having one single point to resolve the DNS name and mitigate risk to having single point of failure.

Also in unlikely case of WATM failure, you have ability redirect your CNAME from *.trafficmanager.net to servicenameN.cloudapp.net.