Re: [PLUG] [slightly OT] (Mac OS X) Is it me or can't you do this with Linux?

[Matthew Rosewarne <mukidohime@case.edu>]

On Tuesday 10 July 2007, W. Chris Shank wrote:
> 2) January will come fast, faster than you would ever be able to switch
> OSes and infrastructure. I recommend ordering as many XP Pro licenses as
> you can afford or will need right now. That should get your current
> infrastructure through another 18-24 months. You'll need this time to make
> any transition.

This is a very good idea.  No matter what you might wish to happen, any 
migration, be it to Vista, Mac, or Linux, will take much longer than you 
plan.  Hofstadter's law will be in full effect.  It is generally best to do 
it in short but solid steps, rather then blind leaps. since just about 
everything in the migration can be done in gentle stages over a long period.  
Don't set hard completion dates and try to keep the timing flexible.  

> 3) Xandros is probably the most Windows-like OS and will probably work more
> reliably with your AD/Domain infrastructure out of the gate. I don't
> particularly like this distro. I used it years ago, so maybe it's different
> now. But when I used it they had all sorts of weirdness - like rewritting
> critical conf files on each boot and releasing new versions every 6 months
> and requesting payment to upgrade.

Xandros is an unwise choice of distribution, unless it's purely being used as 
a transitional stopgap measure.  The distribution's prospects are quite 
bleak, and I would not want to bet my infrastructure on a company that has a 
such a high chance of disappearing.

> 4) Ubuntu is probably the best FREE linux for desktop use, however I think
> it lacks enterprise management tools. You can still manage it in bulk - but
> it takes a lot of know-how and custom scripting to do this. SuSE may have
> good enterprise tools at this point and if Red Hat has their Red Carpet
> service working for desktops these should be considered.

Most distributions lack good enterprise management tools.  It will take a long 
while for your staff to figure out what works well for them, so any serious 
rollout should not be set in stone until you get a comfortable setup.  
Depending on what you plan on doing, Red Hat and Debian are probably worth 
considering first.  (more on that below)

> If you are SERIOUSLY considering a switch - here is how I'd do it:
>
> A) Get senior management buy-in 100%. Without this, all else is doomed to
> fail.

While getting full approval by the senior management is an absolute must, try 
to get as many others on board as possible.  Explain that not everything will 
be perfect, but after the rough spots are polished the effort will pay off in 
spades.  Give the actual users an impression that they have a stake in it 
too, and that it is for their benefit.  Ask them what they would really like 
to see in the new system and keep their feedback in mind.  Also make sure all 
of your techies are enthusiastic about this, since the migration will 
probably require a lot of effort from them.  The more people who are hopeful 
rather than sceptical, the easier this will be for you.

> B) Convert Windows NT/AD domain to Linux Samba/OpenLDAP infrastructure as
> much as possible. This will give you a good non-proprietary LDAP for both
> Windows PCs and Linux Desktops to authenticate against. If you must have
> windows servers, they should be able to join the Samba domain as well. You
> may need at least one Windows Terminal or Citrix Server for those pesky
> windows apps that you can't live without.

Obviously, the backend must be in place and working before the front end.  
Another important aspect is file formats; try to roll out OpenOffice and 
other cross-platform applications on the Windows machines before a move to 
Linux.  Another options besides using a terminal service to run those old 
windows applications is wine, or it's suit-clad sibling Crossover Office.  I 
have had a great deal of success with both of these.

> C) Develop a "Golden Image" of your Linux desktop with all the
> applications, codecs, etc you need. Setup your terminal server / citrix
> clients to run the applications directly. I forget what this is called, but
> essentially you are running the app on the terminal server and only the
> app's window is on your client. It give the illusion the app is running on
> the local PC. Your linux desktop should be setup wth automount on the users
> HOME and other Windows shares. This part takes a lot of thought because a
> user may need to save files from Linux and open them with the Windows TS.
> This needs to be pretty transparent or users will revolt.

Depending on your setup, you may opt to go for the thin client approach or the 
full-desktop approach.  For a desktop approach, use the 
preseeding/auto-install capabilities available in most major distributions.  
Set up your own local package repository and have your machines set to 
upgrade from that automatically.  For a thin-client approach, I would 
recommend LTSP (the Linux Terminal Server Project), which is the most 
developed thin client setup available for Linux.  What approach you decide to 
take depends on your existing infrastructure, the needs of your users, and 
your plans for the future.

> D) Move single task users or other "early adopters" to the new image so
> they can test it. It's critical that these users are either eager to make
> the Linux desktop work or are so clueless that they won't know the
> difference. Any users moved reluctantly at this point will only derail the
> effort. They will find NO positives and needle in on every negative they
> can. Trust me.

The first person to get onto the new system should be you, or whoever is in 
charge of teaching the other techies.  Then the techies must all use it and 
become familiar.  Consider a certification from the Linux Professional 
Institute (ignore Linux+).  Once the techies are comfortable, hawk your new 
setup to the few curious early adopters, and make sure you give them 
something to take home to their own computer.  Talk with these people often 
to assess their impression of the setup and make them feel appreciated.  The 
next victims will be those who only use the most basic functionality of their 
computer (web, mail, documents, etc).  They are relatively easy, since all 
their needs are perfectly served by any platform.  The harder next step is to 
migrate the more advanced users who may have special applications they are 
fond of or have heavily customised their setup.  Make sure you meet with 
these people _in person_ and help them through the process.  Many of these 
types will not be pleased with the change (especially at first), and will 
require significantly more backwards-compatibility labour.

> E) Go to C - rinse and repeat. With each repeat you should begin to win
> converts.

Unfortunately, you cannot rely on too many people actively volunteering to 
switch.  Humans are stubborn creatures and tend not to try new things.  Once 
you exhaust your supply of volunteers, you have to start pushing.  Again, 
take it slow, in particular with the more advanced users.  Don't expect to be 
entirely rid of Windows for quite a long time, but keep pushing it ever 
closer to the side.

> F) Gotcha's and pitfalls. Watch out for sophisticated outlook users. I've
> not yet found a Linux email client that makes them happy.

Sometimes people just need Outlook, or MS Office, or Photoshop, etc.  Make 
sure you allow them to continue using their existing tools if you don't have 
an acceptable alternative.  Pushing too hard on these people will derail the 
migration, so use kid gloves.  However, if there is a viable alternative to 
the tool these people cling to, make sure you provide them with training and 
hand-holding to make the switch.  They will become the aces in your sleeve.

> Note:
> What to do if your infrastructure runs Windows AD, Exchange, and SQL
> Server:
>
> Plan to migrate to Vista and look for a new job.

1) Find the fire exit.
2) Run screaming.


On Monday 09 July 2007, Art Alexion wrote:
> I work in a windows shop. We have been informed by our PC vendor that only
> Vista will be available as of January. As IS and the users will all be
> forced into a change, we decided to explore non-MS alternatives, including
> Linux distros with windows-looking interfaces and Mac OS.

I would not recommend Mac OS as an alternative to Windows.  It is seriously 
lacking as a business platform.

> And, what suggestions for a Linux desktop distro that minimizes the shock
> to reluctant converts from windows?

Trying to find a distribution that mimics Windows is a losing game.  If users 
expect the new setup to be just like Windows, then they will be annoyed every 
time it differs.  Instead it is a far better idea to go for a system that 
works well, is reliable, has a strong future outlook, follows the mainstream 
of FOSS development.  Novell's products have been problematic in a number of 
aspects, and their organisation shows a fundamental lack of understanding of 
what they're actually doing.  Xandros and Linspire are niche products with 
highly uncertain futures.  While many people get feisty about Ubuntu, I keep 
running into problems with its quality and reliability whenever I try it.  
Just like it is said that "nobody ever got fired for using IBM", the safest 
pick for an enterprise rollout is still probably Red Hat.  They have 
incredible experience with FOSS and much of what they develop gets used 
throughout the Linux universe.  Being a Debian user myself, I would also 
recommend Debian.  The Debian development process is overwhelmingly rigorous, 
which results in a distribution of unparallelled stability and flexibility.  
Commercial support for Debian can be obtained from a multitude of third 
parties, whereas support for Red Hat would most likely come from Red Hat.  
Both distributions are solid and capable, your choice would depend primarily 
on your own personal preference and your organisational goals.


To recap, there are a few major points to permanently burn into your brain:
1.  Take it slow, do it right the first time.
2, Make sure you are ready _beforehand_, make sure your techies are ready.
3. Get agreement from as many people as you can.
4. Ask for help, chances are someone else knows how to do it.
5. Watch out for the advanced users.

Lastly, if things don't turn out well, or if the plan is scuttled before it 
starts, don't sweat it.  FOSS isn't ready for everyone yet, so don't harm 
your business or your career by forcing it if this turns out to be true for 
your situation.

Attachment: signature.asc
Description: This is a digitally signed message part.

___________________________________________________________________________
Philadelphia Linux Users Group         --        http://www.phillylinux.org
Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
General Discussion  --   http://lists.phillylinux.org/mailman/listinfo/plug