sanjay nadkarni (Laptop) wrote: > Virginia Wray wrote: >> Hi - >> >> I've been thinking more about the problem statement. I would like to >> broaden and hone it at the same time. Here is what I've come up with, >> and I welcome any feedback: >> >> >> "To provide a mechanism for delivering the AI manifest and install >> image to the AI client which is scalable, delivers high performance, >> and provides a good user experience." >> >> >> >> I've used the term mechanism instead of webserver so that we leave >> options open. I think "delivers high performance" is a little vague, >> but I think that can be solved by defining what high performance means. >> >> I was talking with Sanjay, and he asked me a couple of questions - >> >> 1. For what we are doing, is it worth having an entire webserver to do >> this? >> >> 2. In the user environment, is the data that we are dealing with going >> to be static or dynamic? >> > To elaborate, can we clearly describe what types of data we will be > dealing with this mechanism. We probably have a good understand of what > we do now. But what about in the future ? >
Types, volumes, and peak loads would all seem to be important characteristics to define. And I'd actually throw one additional requirement in: that the mechanism be modular with respect to the rest of the installation, such that it can be evolved/replaced with alternate protocol sets. At least one user has expressed a lot of interest in using torrents, for example. Dave
