Extending TUNe for Autonomous Management of QoS at Application and Network Levels

This paper presents a solution to respond to the increasing management needs of complex networks and applications in datacenters. It presents a way to describe QoS needs at both application and hardware platform levels within an autonomous management tool called TUNe. High level graphical languages are introduced to describe autonomous management policies. The provided extension allows dynamic reconfiguration of both application QoS profiles and network hardware QoS classes depending on the global environment state. Experiments are being conducted with a P2P system called Azureus on a set of Cisco 7600 routers.