James, Nice, thoughtful, post. Thanks. I think that you very accurately summarize the current situation...and the thinking that got us to where we are.
James C. Cotillier wrote: <snip> > OS install experiences typically are targeted to the greatest common > denominator in technical expertise, across the range of their likely users. > But these ranges differ: the typical target installer of Windows likely > has different requirements, goals and levels of technical expertise than > the UNIX sysadmin. Absolutely true. For now, we're certainly not competing with the Windows market for most cases, and we're expecting a more technically savvy audience. > > To suggest that the overall quality of Solaris could in some way be judged > by the snazziness of its install GUI seems to me equivalent to suggesting > that the overall quality of a Boeing 767 can be determined by the available > upholstery color combinations on the flight deck. > Yes, that's true as well. That said, there are other factors to consider here. Linux is a significant factor, and is good enough for many audiences much of the time, and for that reason, we need to consider factors other than just the underlying technical quality. Let me morph your analogy into cars--I think that you'd probably agree that you couldn't judge the quality of a car based on the upholstery combinations either. For the people who want/need a specific model, that's fine. (E.g., a buyer for a fleet that has standardized on Ford trucks doesn't really care about the upholstery options available.) On the other hand, for the people who just want to get from point A to point B, and consider cars to be largely commodities, the upholstery could certainly be a differentiating factor. A truck purist could well have strong opinions on the technical merits of a Dodge with a Cummins TurboDiesel vs. a Ford with some other engine...but those differences or technical merit aren't relevant for many of the buyers out there. That's why you can get spiffy leather interiors on trucks, even when that's arguably not needed. > If you are installing Solaris, then most likely you have selected Solaris > for the application, based on your judgement from prior knowledge and > experience in operating systems. We don't need to market too much at that > point. I think we can assume that you know what Solaris is, that you know > what you are doing, and that you have consciously decided on Solaris over, > say, a Linux distro, for your own specific and technical reasons. > Yes, that's certainly been the case in the past...but that's why Solaris is a niche player in a large market. Sun's historic inability to look past the technical merits of technologies to envision how to connect to customer's business needs is firmly rooted in the assumptions you outline above. I hold that we need to change, and quickly, to a far more customer- focused model. No, Solaris won't ever be the laptop OS of choice for the masses, nor should it be. But it shouldn't also come with such barriers to adoption that it's not viable for anyone who isn't already an expert. > I hold that this suggests that you don't need to be splashfully entertained > a whole lot while installing, and moreover, that such antics may even be > [i]counterproductive[/i] to a trouble-free install. You most likely know > exactly what you want done for the install, what to put where, and what > configurations to set. Any CPU/graphical engine cycle not devoted to > [i]install action[/i] is a candidate for a wasted cycle. > For some audiences, that's true. But for many other audiences, an aesthetically pleasing installation can be significant. Given the power of current systems, I think that we can afford to waste a few computing cycles to provide a better experience for those who want or need it. > Have a GUI environment at all? Sure--no intention here to bash GUI's in > general--you can open up an xterm and do some local stuff while the install > proceeds, etc. (just being careful not to step on feet). > > As far as unabashed marketing during our install--did someone say "popup > ads"?--what a way to treat a masterpiece--one might as well try to market > the latest in sunglasses by putting a pair on Michelangelo's [i]David[/I]. > [Analogy backfire--actually they'd probably sell millions due to schlock > shock alone...] > > Summary: > > Make the install experience efficient but robust, clean, and most > importantly, professional and reserved in appearance, befitting a system of > this quality. Agreed--I don't think we need dancing paperclips or the like. > > Assume the installer knows what s/he is doing: allow overrides and > do-it-anyway-because-I-say-so's as much as possible. > For some installation contexts, yes. But we should also provide a clean and quick path through the install. > When in doubt, understate rather than overstate. Your > installer/user/sysadmin has already been sold on the product. > Doesn't this doom us to perpetually being a niche OS, and keep us with only the same audience and customers that we currently have? I think that Solaris offers some really compelling advantages over Linux (and Windows) for many markets, but those advantages aren't significant enough to dislodge the momentum behind Linux when Linux is good enough for the current needs. For example, I established a couple of servers at a colo facility some 10 years ago when I was self-employed. They were running Linux, and that's persisted to this day. No, it's not the best solution from a technical standpoint, but it's good enough, and was quick and easy to install. There wasn't ever a compelling reason to change. Even now, the only reason that I'm pushing Solaris out on those servers is because I have a strongly vested interest in it--Linux is still good enough for light Web & email traffic for my friends and family, or for most small business applications. I'd encourage you (and everyone) to read "The Innovator's Dilemma", by Clayton Christensen. Among other things, it discusses the issues associated with multiple products that are "good enough", and what's needed to effectively differentiate and win market share. Given that, despite the advantages of Solaris, Linux meets most needs most of the time, we have to find other ways to gain share. From page 218 of my edition, "... when two or more vendors improve to the point that they more than satisfy the reliability demanded by the market, the basis of competition shifts to convenience." While we should certainly continue to provide the enterprise-grade capabilities and flexibility in installation, I believe that we should _also_ be providing the convenience and ease of use to make Solaris approachable to audiences that might otherwise not try it. Eric -- Eric J. Ray Software Engineering Manager Solaris Install Sun Microsystems 303-223-7843 (direct)/x81067 eric.ray at sun.com
