My Experience and Strategy to Migrate MS Windows to Linux

This article is made with a purpose of
giving a hint on strategy for migrating MS Windows to Linux,
specifically using OpenOffice, Samba file server, and dosemu (DOS
emulator).

Background: Why spending hundred
millions of rupiahs (tens of thousand dollars) when there’s another
solution in which much more economical?

It is high time for management to
realize that in the more and more hard business competition era, all
form of efficency must be done. And as we know, in the last couple of
months, Microsoft Indonesia and BSA is getting more active in
pursuing those companies that use their software without proper
license.

Well, it is understandable and we
cannot blame them for being such predator, because they are the owner
of the copyright. However, as we know, the licensing cost of MS
Office basic (only covers Word, Excel, and Powerpoint) is about $150.
So, for a company that owns about 100 PC, it must pay $1,500. With
additional $140 for the Operating System (WinXP). The total
cost therefore is about $300 per PC. Multiply this by 100 PC means
around $30,000. For companies that own around 100-300 PC the
total cost for licensing could be hundreds of thousand dollars!
(Billions of rupiahs).

As much as we want to migrate, it is
not an easy task to do. Company still should consider the cost and
benefit of migration plan. Therefore, the strategy in general should
cover these:

1. Do an analysis of PC usage.
   We must inventory what programs are used, which PC is used for
   administratif task only, such as word processor, spreadsheet, etc.
   Which PC is used for certain program, and on which platform and
   programming language it is made, etc.

2. Based on the data we gather, we
   then analize deployment coverage. In this analysis, we
   research on possibility of all migration track:

– Which PC that
can be 100% migrated, including the OS and the office application.

– Which PC that
can only be 50% migrated, is it only the office application?

– Which PC that
cannot be migrated at all, in other words 0% migratable.

3. Based on the above mentioned
   scenario, we then also analize the cost needed to be
   presented to the management. Of course the ideal plan is to migrate
   all the PC and applications, but, from my experience this is usually
   not possible yet. Because usually some of the function of the
   company has already been bound by Microsoft/other propietary
   products. It is a sad fact indeed.

Case study:

To give a clearer picture, let’s
discuss the following case.

A mid-size ad agency has about 300 PC
that is located in several branches, in and out of town.

Lately, the management is worried
because it has received a letter from Microsoft Indonesia, urging
them to do software audit and the status of the license on the PC.
After looking from the industry, it is known that the license cost
for WinXP is around $140 per PC, and for MS Office basic is around
$150 per PC. So the total cost for licensing all those 300 PC will be
around $90,000 plus another thousand of dollars for servers
and its CALs.

Now, with the above mentioned analysis
steps, the management gets the following data:

1. Around 30% of the PC only run
   office application, that is word processor and spreadsheet.

2. Around 50% of the PC run an
   accounting and sales program that is written in Visual Basic and
   shared in a MS Windows server.

3. Around 20% of the PC run a finance
   program that is written in Clipper and shared on a Novell server.

4. The specification of the PC is
   varied, from Pentium classic (mostly for clipper application),
   Pentium 2, 3, and 4.

Ok, now we can start focussing on what
we can migrate. Because of the limited time, then we must concentrate
on PC that can truely migrated without many obvious problem.

Let’s left behind those PC that run VB
application for the time being.

For PC that only run office
application, we can migrate them right away, but because of the
varied hardware specification, we must make a detail plan as follows:

1. For Pentium Classic, we use Mandrake
9.0 or RH 8.

2. For Pentium 2 and 3, we use Mandrake
9.2, 10.0 or RH 9.

3. For Pentium 4, we can use the latest
distro such as FC4 or Mandriva 10.2.

For PC that run clipper application, we
can simulate it using Samba file server and dosemu.

This is the trick:

1. We copy the application into a
   Samba server share.

2. Make the usual share configuration
   with added configuration:

– Create the same
user in the server and in the workstation.

– chmod 777 the
share directory.

– add these lines
into smb.conf:

create mask = 777

default case =
upper

preserve case = no

3. chmod 4555 /usr/sbin/smbmnt in the
   workstation. This will assure that common user can mount the share.

4. Add this line into /etc/fstab on
   each workstation:

//ipofserver/sharename
/home/user/mountpoint smbfs rw,user,noauto

5. Install dosemu on each
   workstation.

Then we can connect to the Samba
server, and run the clipper application this way:

1. Open a console in the workstation.

2. Connect to the samba share and at
   the same time mount the mount point:

– mount
/home/user/mountpoint

– enter the
password when asked.

3. Run dosemu in that console.

4. Enter to the mount point (default
   on drive D), and run the clipper program as usual. Done! 🙂

From the above explanation there is
hope that we can make a substantial efficiency on licensing cost.
But, from my experience, there are still more analysis to be done to
make sure the success of the migration plan, that is:

1. More comprehensive test
   must be done on the compability of the dosemu environment, samba,
   and those clipper programs. Because as good as dosemu can imitate
   DOS environment, it is still not a native DOS, and I cannot
   guarantee that is it 100% compatible with various clipper programs.

2. We must also research and test
   more on the office application used by the users. Whether there
   are features that are really MS Office specific, that cannot be
   imported into OpenOffice in good shape. The latest OpenOffice 2 beta
   1 is really-really good in compability with MS Office, but still it
   is not 100% compatible. If we encounter this incompatibility, we
   must then decide whether it can be overcome if we remake the file
   natively in OpenOffice, or cannot be done entirely.

3. And another important task is to
   backup all the files that will be migrated. So if there is a
   problem, user still has the original files. I have encounter a
   situation when a user can import an excel file just fine into
   OOcalc, but then after several open and save operation, OOcalc
   failed to open the file.

4. The last thing to consider to be
   done, and maybe the most important success key, is to give
   understanding to users why we migrate. Give an explanation that
   is acceptable, so that we will have user support on the migration.
   Be there when users need help on using the new program and
   environment. Give an easy comparison on the menus, such as in MS
   Excel we open this menu and this menu to print, and in OOcalc, we
   open this menu and this menu. So that the users can see the
   similarity or the difference clearly.

Well, that is the strategy I can share
or to be more exact is my experience in migrating MS Windows
environment to OpenSource solution specifically GNU/Linux.

Is short:

Why spending hundreds/billions of
rupiahs if there is a much more logic solution?

Additional info:

Although done in low profile and not
known by the general public, some big companies in Indonesia are
migrationg their PC into Linux. So, silently, OpenSource movement is
being adopted and growing in Indonesia.

I hope this writing can give an
inspiration or a general picture for us who wants to grab a logical
solution on licensing cost. It is indeed a high time for management
to migrate to Linux, as it is a logic, feasible, and proper thing to
do for any management that cares on the business.

v.1.0 by ari_stress a.k.a tiger74 a.k.a
Fajar Priyanto
Jakarta, 17 September 2005. fajarpri at
arinet dot org
He is a Microsoft Certified
Professional who falls in love with Linux. Works for an automotive
group in Jakarta.