UNIX admin wrote: >> Other than simply disagreeing, can you provide some reasons why >> this layout would unacceptable to users? > > I have already done so, did you not read what I wrote? > > That layout does not consider OpenSolaris target audience at all. > Your target audience knows nothing of LU, and they're stomped with > the concept of slices inside of a BIOS partition. > > And: that layout wastes 15GB of space for something that in most > cases will never be used! > > LU is for sysadmins, not end users. In case it slipped "through the > cracks", most people are afraid of the CLI and have difficulty using > it. And even if there was a GUI, the concept of LU is so "off the > wall", your end users would be stomped anyway. >
I'm not sure I understand how you are intending to differentiate "end users" vs. "sysadmins" but I'll assume the former are desktop/laptop users and the latter are server administrators in corporate environments and play it that way. I agree that LU is not something the average user will know about up front, and I will agree heartily that it has a number of usability issues. But I will strongly disagree as to who will find it useful. Once they've found out about LU, we find more "end users" taking advantage of it than the "sysadmins". "sysadmins" at many environments have standardized procedures (or, in some cases, superstitions) that they've developed over the years and are leery of trying out something that seems so radical. On the other hand, the average users say "I can upgrade without downtime and without worrying about ending up with a brick? Sign me up!" with little prompting, once they are educated about it. The main issue both face, however, is that if they've already carefully laid out their disks and didn't leave slices available, then making the change is *very* disruptive and far less likely to occur. So if we want people to adopt it, we need to make it the default. > Finally, any good sysadmin will know about LU, and will know how to > slice up the disk in just about any way imaginable. That audience > definitely does not need presliced and preinstalled systems, becase > they will know what they want and "one size fits all" factory layout > certainly won't work! > Your definition of "good" is probably a bit more stringent than the marketplace we'd like to be using OpenSolaris; "experienced in Solaris", perhaps, but even there you'd be surprised how many don't know about LU or have never tried it. > In other words: it's a perfect example of vendor believing that he > knows what his customers want, and at the same time not listening to > his customers. > Well, we have talked to many customers (both existing and potential), and you will undoubtedly be disappointed to hear that most of them disagree with you. I've been saying we were going to do this for 18+ months, even before the original installation strategy paper was published to the OpenSolaris community. The feedback has been overwhelmingly in support. Dave
