Virtual Network functions leveraged by Telco and enterprise demands 5 9’s of availability (5.26 Minutes downtime per year) and also higher data plane throughput per VM.
SR-IOV is commonly used method to support maximum possible throughput by the Server and Network interface card in the Server. SR-IOV bypass the hypervisor and takes away the Live and offline migration features which can be used during server failures to move VM to another server.
In addition; AT&T used several tricks of the trade to achieve higher availability ( 5 9’s) of VNFs on 3 9’s OpenStack cloud instance such as automated placement of Virtual functions, Control loop fault detection and auto recovery, etc.,
In this talk, we will share all the methods we have tried so far to achieve higher reliability of virtual network functions using one of the large OpenStack cloud deployment in the world.
Attendees will learn all the methods AT&T has leveraged to run highly reliable virtual network functions on top of OpenStack Cloud.