Warning ! You can put GNU/Linux at risk

     Ok, finally after lots of advocacy from you (and some help from the press), your boss decided to let you try GNU/Linux in some minor tasks, just to try it out.

     You are happy and run for the CD that you always carry in your lunch bag, or that is hidden in your desk, waiting for this moment.

     But I have to warn you, my friend, this could be one of the worst mistakes in your career, and you can put GNU/Linux at risk if you are not careful.

     Rolling out GNU/ Linux in a big company is not an easy task, specially if you are trying to do this alone. Even worse, if your company is Windows-only, and you are the only soul that knows something about Unix. Hey!, even if your company is a mix of Unix and Windows, you have to watch your step.

     If you are planning to start using GNU/Linux as the file/print server of the 5 or 10 guys at your office, it is not a big deal. You get a hold of an old machine, install your favourite distribution, configure Samba, configure the printers, and you are done. BUT!!!, but!!, if you are trying to roll it as a desktop, be very, very careful!!!


     Well, first because you have to choose the correct persons to be the testers. You can not just go like a tornado erasing Windows and installing GNU/Linux in everybody's PCs.

     First install it on your PC, be sure that you can make it work! This includes all the hardware that is included in the "official PCs". If you encounter some hardware that is still not supported, try to find a workaround or try to find other advantages of GNU/Linux that overcome this hardware issues. This could be very easy, or very difficult depending on the hardware in question. Don't forget to send e-mail to the company producing this hardware asking for support. There are also some companies that could develop a driver for you, this of course needs some budget. As a last resource, start coding yourself the device driver or try to get in touch with some developers and maybe they will accept to create the driver for you in exchange of a piece of the hardware. There are also a couple of companies out there that match developers with companies for special programming needs.

     Ok, the hardware is running fine, now let's go to the software. It is nearly impossible (until today, December 1999) to find the exact same software that your partners are using, available for Linux. So, you have to look for substitutes, and this part is very important, get a list of the features of the software they are using and then look for a program or a group of programs that fulfill this needs on GNU/Linux.

     If you cannot find all the features, at least look for the one that is closest to this goal and try to look at other features, like being Free Software and all the benefits that this brings, like customization, etc.

     Now you have hardware and software running in your PC, and you are able to do your everyday work completely from GNU/Linux.........until somebody sends you a document in a proprietary format (.DOC, .XLS, .PPT, etc.), you find some filters, or some software that can read this kind of document (StarOffice, etc), but this is a very sensitive issue, if the rest of your company (let's say more than 200 employees) are going to be sending this kind of documents back and forth, and these are the "official formats", and your software can't give you the needed percentage of compatibility, then you will need to go for an emulator or virtual machine software (up to this writing, the only commercial is VMWare, but there are others on the work, including one that is Free Software, Freemware).

     Some of you could say: "Hey, but I use 'strings' as my Word filter", yes, that could work for you, but when you are thinking in rolling out PCs for everybody you have to think that maybe this is not going to be good enough for them, people can be very picky!!. This is why you need "outsiders" to test your Linux configuration, because after a while you could lose sight of some details that you take for granted, like mounting the floppy before using it, etc. But you have to know when to start recruiting outsiders.

     After some time, maybe months of running this way, you can start considering rolling out this to some other people. Some of you could think that it is better to start the whole team, or at least a small amount of guys chewing on it, so all of you can research on one aspect of all the issues I said, and that way you could accomplish it faster. But, here is the danger!!!

Involving more people (a.k.a Getting guinea pigs)

     If you involve more people, that means that more people is "spending" time in dealing with GNU/Linux issues than doing "real work". Some of your partners, or even your boss could object to this, saying that you are only "wasting" time. It happened to me, and can happen to you.

     If you want to involve some other people in this testing, either at the beginning or after some months of you running GNU/Linux in your PC, please take in consideration the following issues when "selecting" who are going to be your "guinea pigs".

     First, find out why they want to participate in this project, and choose the ones that are REALLY interested, not the ones that just heard that "Linux is very cool" so they want to know more about it. How to know who is REALLY interested, well, for this you need a little of psychology, talk with them, show them the installation you have and wait for their questions. The type of questions they ask you tells a lot about their interest, and their intentions.

Then find something about their background, it is better to "accept" a Unix guy than a Windows guy, at least for this initial stage. If you have plenty of Unix guys, it is better to choose one that has system administration skills than one that has been a "mortal" user all the times he has been in front of a Unix Workstation.

     Another important aspect you have to consider is their available time. It is not a good idea to "enroll" the busiest person in the office. But be careful, even the busiest one could create some spare time for this project (lunch time?). So basically this is a question of disposition, remember that the excuse: "I don't have time" is not real. A person that is really interested will find some time, maybe doing some installations and configurations at home, and applying them for 10 minutes at work, or at least letting you know about them via e-mail.

     Experience inside the company is also one important factor, you would prefer a more experienced person inside the company to join your project, mainly because this person has already some credibility, so if he/she says that "this stuff is good", more people are going to believe him/her than the new kid on the block. If you are this person (the experienced one), then you could choose then a new guy, so you can shape him/her by providing a tutoring image to follow, helping GNU/Linux to go deeper into his/her mind (brainwash??).

     Now, I will go deeper into some of the things I already said.

Getting deeper

     If your office is Windows-only, and nobody has experience with Unix, I insist that you have to wait until you have been working from you GNU/Linux PC for some months, that way you can do an extensive testing of everything, hardware, software, etc. And they can see some results before they even try it. A good idea is that if they can connect to your machine via XDM or GDM then you can open an account for the interested and let him play around your machine, this has the disadvantage that your machine load is going to be splitted among the users, but the good thing is that you only maintain one system instead of various. You should ask them to report you bugs, or features that they want to have, that way you polish the system for the roll-out.

     If your machine gets too bogged down because so many guys wanted to use it, then you could get a hold on another old machine and build a "user server", so they connect to that machine instead of yours. Just be sure you keep always that machine updated with the latest versions of all the software, at least the same versions that you are running in your PC.

     During this testing time, try to put GNU/Linux in a strong position in other aspects (just in case), like system/network monitoring, file/print server, etc. Something that will stay even if the desktop roll-out is aborted for any reason. With this at least you will have something running GNU/Linux there, and at least your GNU/Linux desktop could be justified.

     You should avoid Windows-only people at the beginning because they tend to think "The Microsoft Way", and try to do the tasks in the same way they do it in Windows, just because they haven't experienced other operating systems, so they can become frustrated very easily and will say bad thinks about GNU/Linux, just because of ignorance, specially if they are not REALLY convinced, and they just want to "play with the new toy, to see what it is like". On the other hand, people that has been using only Microsoft's operating system and are fed up with them, and they are REALLY conviced that Microsoft is not the latest/greatest, can be very helpful, but if you want to stay on the safe lane, stay away from Windows-only people.

Linux-newbies under your wing

     Having "Linux-newbies" under your wing could be very stressful, specially because they are going to call you for issues like:

You might say:

     "Stressful?, Hugo says that this is stressful?, he should take a look at the end of the month when I have to pay the bills!!"

     Well, the stress comes when you have all these Windows-only guys/gals doing their first steps on GNU/Linux and showing off to the rest of the company, and then the invevitable happens!!!!, they forget to do the 'export DISPLAY=:0.0' when doing a 'su -' in order to open an account for their partner with Linuxconf, and they start struggling to make this happen, making look himself fool, and GNU/Linux as a "not friendly" environment before the eyes of the person they are showing off this.

     The result: your apprentice looks like a fool, the guy that was waiting for his account thinks that either GNU/Linux is not good enough, or that it is "too hard", finally giving a "bad score" for you and GNU/Linux.

     Another bit of stress comes when your apprentice is talking about "this new stuff that I am testing", and the gal that is talking to him asks if there is a software to do this, or that, or if he can show her how to do a backup of your home directory. You apprentice starts trembling, trying to find an answer, to finally say: "No, I think this is not supported yet", or something like that!!!

     Remember, it is better to say: "I don't know, but give me some time, and I will do some research about it to find the answer to your question". Of course, if it is REALLY not supported, you SHOULD NOT LIE!!!, always say the truth about what GNU/Linux can or cannot do.

     Another thing you have to think when selecting testers is what kind of tester you want: the one that keeps bugging you because they don't have the "Windings" font installed or the one that helps you fixing things?

     If you have a lot of testers that only ask you for features to be implemented and they don't take time to learn, configure and/or understand, your support bandwidth will be consumed very rapidly and you are going to be fixing little details (colors, fonts, etc.) instead of concentrating in the big picture (evaluating applications, network interaction, system configuration, task automation, special scripts, etc.).

     One last advice, remember that YOU suggested GNU/Linux, so you are putting your "professional image" at risk too!!, that's why it is important that you REALLY know GNU/Linux, if you don't feel so sure about this, go to your local GNU/Linux users group meetings and keep practicing at home before letting a bad impression of GNU/Linux and yourself in front of everybody at work.

     Well, I think that it is already a long article, and I hope I put some light in your way for a good Linux advocacy. Not everything was covered, but typing on a laptop is not good for my wrists, so maybe later I will continue with more of this kind of articles, all depends on the feedback I receive, so start flaming me, or at least say: "Good article Hugo!"

Yours Truly,

Hugo Gayosso

December 1999


     All what I have expressed here is a product of my own experience in introducing GNU/Linux in big companies (SIEMENS, GM), and in helping others to introduce it to some other companies. I am not trying to stereotype anybody, any system administrator profile, etc. I am just talking about what I have seen.

     "What are your credentials???", well, I think that I succesfully introduced GNU/Linux in a manufacturing plant of SIEMENS, doing all sort of tasks, from servers and terminals in the production line, to workstations for engineers and system administrators, and I say that it was succesfull because even that I am not working there anymore, GNU/Linux is still being used, and not only that, its usage has been increased, and now (according to my ex-partners) all the new projects that are developed there, are being done on GNU/Linux. About, General Motors, well, this is an ongoing project, up to now I have been able to place some GNU/Linux machines, doing some monitoring stuff, and some other tasks, not any production tasks yet, though. :(

Copyright 1999 Hugo Gayosso

     The copy and reproduction of this article on its entirety is permitted as long as this copyright notice is kept, and no modifications to the article are performed.