Well, the only real way to "abolish" the scalability problem is to insist on solutions that have zero growth. That's not necessarily feasible or a requirement. After all, hardware is getting faster and we can reasonably take advantage of that. However, even with a zero growth solution for 3.1, it's trivial to imagine a solution for TE that still results in non-scalable growth: for example, generate arbitrary more-specifics, down to the host level.
The purpose of all of these design goals is to help us agree on the criteria that we will use to evaluate various proposals. We are trying to establish the metrics that we will use to judge proposals on, but are also trying to leave open the possibility of using our judgement to make intelligent trade-offs between various metrics. Thus, we are trying to avoid creating arbitrary requirements on those metrics. Tony |