Most research in the QoS specification deal only with QoS specification at the application level and had little to do with the SLA to QoS mapping. People have come up with langauges with QML, HQML, CQML etc which specifies QoS at the application level and then the applications above interact to ensure that the QoS is met or not. The following example tries to make this clear.
Say, a client which has its QoS requirements specified in QML requests for a particular service from a server which may or may not be capable of providing it the required service. Thus, they interact say in the CORBA environment and fulfil the request.These have little to do with the co-relation with the network performance metrics.
[LSE03] comes with a language SLAng, a language for SLA (service level agreements) which describes the proper syntax for SLA. These people say that SLA can be of 2 kinds : vertical and horizontal. Horizontal SLAs are the ones at one of the followinglevels:
1. Application level.
2. Container level.
3. Network level.
And then there are vertical SLAs which maps the SLAs vertically which clearly is also our objective. But, he just came with the way to represent SLAs starting from SLS (Service level specification) of the Tequila engineering work.