From: Darryl Bond
I read your Viewpoint regarding the impact of Linux on large organisations (ComputerWorld, October 2, p24) and would like to offer a few points.
I am the IT support specialist at a large power station in Queensland. We have been using Linux for four years in a "production environment". This was initially to get reliable PC X terminals for a new control system based on Solaris. Since then the use of the Linux operating system has branched out into other critical areas.
I wonder why you conclude that the use of Linux will be successful only when an organisation has abandoned all flavours of proprietary operating systems and gone for Linux? This misses the point of Linux entirely. My major liking of Linux is its ability to integrate with other operating systems seamlessly. The user on Linux can use Unix/Windows applications without knowing what has occurred behind the scenes (X Windows, Citrix).
Some examples of the integrating qualities are: NFS access to Unix file systems; SMB access to NT file systems; serving both the above; NIS/NIS+ authentication; NT authentication; ONC RPC services (Source code compatible with Solaris); standard scripting languages (perl,TCL/Tk, Python, etc); reliable, high-performance networking stack.
I see the success of Linux as the same as proprietary operating systems (Unix and NT etc) where the right job is done with the right tools for the right price, reliably. Reliability is important and Linux reliability has been excellent.
Support of Linux is certainly different (e-mail, newsgroups, HowTos, Linux Doc Project) but when the crunch comes you have to be able to support yourself by what ever means fall to hand.
As an example, our Y2K project has to validate the systems used on site. The open source systems were a breeze. All the code was available for inspection.
Our programming expert was able to validate the date handling of the Linux OS and packages used on site by inspecting the source.
Consider the Windows NT case. More than 12 months ago Microsoft was claiming Y2K compliance. Since then many Y2K issues have come up on MS operating systems and the patches are still not forthcoming. When they arrive the team will still not be able to validate their compliance without a great deal of testing. In the end it comes down to hope that somewhere throughout the world all bugs will be uncovered by stumbling across them and that they will be then resolved.
If an open source package was not compliant, or broken, then there is always the option of fixing it yourself (or paying someone to fix it for you).
All operating systems have their place and I see no need to select one only over all others. It is better to have a mosaic of interconnecting correct systems than one compromised system (support issues not withstanding).
The opinions above are my own and not necessarily my employer's.