Reading along on slashdot yesterday, an article titled "Netcraft Web Server Stats Challenged" caught my attention. The article referred to a report in the Age that a company called Port80 surveyed Fortune 1000 companies and found that more than 50% runs IIS, then Netscape followed by Apache.
This made quite a few people choke on their cup of coffee that morning, witness the large number of posts regarding Port80, their software, their server mask application etc.
Because these results appear to go flat in the face of what the rest of the world appears to be running according to Netcraft (2/3 Apache), I wondered if the survey was conducted by way of a phone-call to the CEO or CIO and if they knew what they were talking about.
First off, let me point out that I am a Debian Linux user, I write web software for a living and I deploy Linux wherever I can. I only run WinNT so I can submit my monthly statement to the taxation department and run an abomination called MYOB to send out invoices - if only I'd continued to write an accounting package when I started my business :-(
I don't like Microsoft, detest their deployment and marketing FUD and generally feel that the world is a better place with Linux and projects like Apache.
That out of the way...
Using a Fortune 1000 list that contains registered domain names from a research project which is a year old, I wrote a little script to query Netcraft's servers overnight and collate the results.
While I concede that my Fortune 1000 list is a year old, while some sites had no registered domains, while there were duplicates and a few strange results, I did get a similar result: IIS 50%, Netscape 19%, Apache 19% - there was even one Fortune 1000 company running WebStar on MacOS (RadioShack) and another running Zope on Linux (SPX). Just for clarity, the source file (the list of domains) to query was a year old, but the servers they were running on that domain are current as of the last time Netcraft looked at them.
The results - while not the same as - appear mostly consistent with the Port80 report.
All of this by way of introduction:
If you figure that the higher a company rates in Fortune 1000, the more impact it has in the world, then impact wise, using OSS, we as a community have a way to go before the message gets out.
What I wondered was this: "What if the your company is running something other than what you think it's running?", alternatively, "How could you get Fortune 1000 companies to consider using OSS?"
While I might consider that running Linux is a good idea, its obvious that not everyone shares my view. I've worked in both heterogeneous and homogeneous environments and while there is an argument for both, I strongly prefer an environment where the tool fits the problem.
I postulate that most Fortune 1000 companies use their "advanced computing resources" for File and Print sharing. This is by no stretch of anyone's imagination a really complicated or taxing task for a server. I further postulate that most IT people stay with what they know. (I know that this is not true for all IT staff, but I'm guessing that in a Fortune 1000 company the incentive to stick with what you know is higher than in small or start-up company.)
I know from bitter experience that most companies have more problems with their technology than they dare to admit. Servers go down more often than the boss would like, productivity is lost more than is liked by the share-holders, costs are higher than are visible on the balance sheet.
So, if I combine the above, we typically see that a company that uses IIS is likely to be running some form of Microsoft File and Print sharing. In a time not too far in the past, this was the domain of Novell. While Novell is worthy of a whole story of its own, lets just continue down the other path for now. The IT staff in a Fortune 1000 company is likely to use and deploy Microsoft software because they (think they) know it.
Also from experience, I know that if you deploy a (Linux based) Samba File and Print sharing solution, you will have less server outages, less infrastructure issues and better performance. While there is a learning curve involved in deploying such a solution, my experience is that once it's up, it's up.
What if some bright spark built a Knoppix or Gnoppix Live CD that booted an existing Microsoft File/Print server and ran it from CD with the settings *exactly* the same as the Microsoft hard disk installed server - lets call it Sambix? This would mean that an IT person in a Fortune 1000 company could play with an existing server set-up with minimal disruption, could try out OSS software and could become more familiar with the offered alternatives. The Sambix server would require less rebooting than their existing server, it would be running Linux, it could also be running Apache and other goodies. For this to work it should be implemented in such a way that the existing server install and its settings are used to put the Sambix server onto the network - same users, groups, share points, services, etc. (I realise that this is no trivial task.)
Of course once that happened, we could then begin to see more Apache/Linux combinations in the Fortune 1000 list, which would be a good thing.
A side effect of this would be that you could start to do the same thing with Knoppix and Gnoppix disks, on workstations. So by stealth you are introducing alternatives into the company.
This is basically using exactly the same tactic that Microsoft used push Novell from the throne. Offer CD's that provide server and services for a limited time, then get the customer hooked, then charge them.
Of course in the OSS world we wouldn't need to charge them - imagine their savings - shareholders happy too!
This is only a rambling from a travelling web-developer, do with it what you will. You can send me email onno at itmaze dot com dot au or post feedback on the Linmagau website.
This article also appeared in linmagau.org