The question always comes up in this environment around performance. Sometimes it's brought up in the terms of how does one manage an end-to-end SLA in a cloud environment or in these complex environments? The forum has done a number of pieces of work in this area both in the publications of specifications.
We've got one on end-to-end SLA management in a cloud environment that describes where the various contact points would be in that. In some areas, you're going to have different technologies, different metrics that would be available. It provides some best practice on how to pick up those specific KPIs and the best way to correlate that information on how to deliver and end-to-end service.
We've also developed a set of performance APIs that basically gives you the ability to collect performance information from several different device types and the ability to pull that into your consuming system, for lack of a better term. I don't like to call the network management systems but you have to have a spot where the performance information from the various aspects are pulled in. At that point, you need to be able to understand what the semantics are around that data, how to be able to aggregate that data into a meaningful way.