Linux is still very much on a growth path and requires a firm, hands on approach. Today's Linux user has to be prepared to do some digging, hunting down updated packages and following mailing lists, newsgroups and web sites to stay abreast of recent fixes and developments. I don't have as much time for this as I'd like (although I do enjoy it), because I have to do work at some point in the day. For those wanting a comfortable, fully-featured environment right now, this is a big issue. I've noted a number of areas where considerable effort was required to customise things to my requirement, and even in those there's a danger that I may give up rather than spending further time to nail the remaining irritations.
Some people may feel I'm being a little ungrateful by carping about the bugs in applications such as AfterStep. After all, it's free and you get what you pay for. Don't get me wrong: I think it's wonderful that developers are prepared to make such herculean efforts for no direct monetary reward, and the results are unarguably impressive. I'm fully aware of the philosophies underlying open source and its growing importance. But the Open Source movement now claims to match commercial products for robustness, functionality and reliability. It is therefore only right to judge both by the same standards. Users deserve to be made aware of the rough edges on some high visibility open source projects, and cannot be blamed if they feel the problems do not justify adoption at this stage.
Linux proves itself daily in many organisations to be an excellent choice of OS for low to mid end Internet, Intranet and departmental servers. As the large database vendors such as Oracle announce their support, it reaches for a place in the enterprise world. I hope and believe it will achieve this aim, but it isn't going to happen without a few new failures along the way. As a friend at Sun maintains, enterprise applications will subject Linux to technical and political stresses it has not yet had to face and it will be found wanting at first. There is a world of difference between running a high activity but low complexity application such as a web server, and humouring the intricacies of a large database management engine.
Here, Linux's own past may work against it. Flame me for saying this, but the greater mass of Linux developers can be in little better position to understand the enterprise than Microsoft. Both are coming up from a desktop-centric, low end world. Neither has much general experience of enterprise needs. I could point to the apparently low interest in providing high availability solutions for Linux as an example (whilst the Linux-HA list has held some stimulating discussions, the group currently lacks the impetus and organisation to produce much in the way of actual code). Meanwhile, Unix vendors are scrambling to integrate availability and resilience features into their core offerings, in a bid to match mainframe standards. (The Linux zealot is free to dismiss this as the rantings of a biased, commercially-tainted traitor. :-)
In mitigation, the Linux folks don't have a desktop monopoly to protect and they're probably more talented than the stock-motivated drones that MS must employ in many groups. One would therefore expect them to release a better product and get there quicker. Certainly, Linux is ahead of NT on a purely technical level.