Don't confuse bad Linux support for bad Linux

David M Williams
Wednesday, 02 July 2008 11:52

Opinion and Analysis

I’ve been integrating company networks for years; I’ve had the dubious pleasure of being involved in many mergers and acquisitions. I’ve gone to companies where they have spoken well of their contracted IT support persons. One company called their guy MacGuyver because of his perceived resourcefulness.

I pursued this company for several days to advise their administrator password before I could begin merging their user and computer accounts when finally an administration lady offered the advice that she didn’t think they had one. Sure enough, I went to the server and wiggled the mouse. A Windows XP login screen greeted me. Not Windows Server but XP. I pressed CTRL-ALT-DEL, left the password box empty, pressed enter and I was in.

There was no domain or security. Users logged in to their own computers with local workstation usernames. The file server, such as it was, had no backups going on. And to make things worse, they used ISP e-mail addresses.
I see situations like this regularly. I could complain about how insecure a Windows environment is, and how Windows does not offer any sort of e-mail server or backup system. Obviously, that would be ludicrous – but I trust you see my point. A bad Windows installation can’t be used to come to the inductive conclusion that Windows is inherently bad. (For that, we will use other arguments!)

Similarly, I would trust intelligent people to see that bad Linux support, and a bad Linux setup is, well, just that. Bad Linux support and a bad Linux environment. It does not in itself mean Linux has to be that way. It does not mean Linux file servers customarily crash. Egad, one would have to expend effort to make a Linux file server which goes down like this.

Yet, what can be done? Should certifications like those offered by Red Hat Linux be advertised more prominently such that organisations know to expect some measurable proof of experience by their support partners? Should each brand of Linux – like Ubuntu, SuSE, others – construct their own Linux certification programs?

Or am I overreacting? Is Linux support more readily at hand than I think? Does the case study really only have the message that Glidepath’s integrator alone was flawed and not that there is a more widespread problem with finding good enterprise-level Linux support around the world?

What do you think?