Slashdot Mirror
Migrating Your Office from Windows to Linux?
bastiji asks: "I work at a mid-sized company, around 50 people and 90% M$ shop (10% being the Sun server doing our backups). Most of my users are using Office 85% of the time with some specialized apps thrown in for good measure. With the upcoming licensing changes from M$ my finance guys are worried about increased spending on even the software that we already own. I've been to told to look for alternatives and I'm asking for your help. How does one begin to do migration from a totally dependent M$ shop to the least expensive options. Are there any examples for mid-sized firms taking this route and any public examples of cost-savings?"
In my experience, it's best to replace your servers with Linux, particularly because end-users tend to be more resistant to change concerning their day-to-day activities. If you're using Windows NT for file and print sharing, then you can easily replace those with Linux and Samba. The user's won't care what the server OS is so long as they have access to their files and printers. Most of your cash outflow can be stopped here. Of course, if you have specialized apps that require MS software (ASP scripts), then you'll need to maintain those, but for pure file and print sharing, you can easily go the Linux/Samba route.
For internet services, you're set. Linux can do everything Windows can and more. For SQL services, you can migrate data from MS SQL to any myriad of free SQL servers available from Linux. Just make sure that your SQL statements are ANSI compliant.
You'll just have to handle these on a case-by-case basis.
For the end users, consider OpenOffice as a replacement for MS Office. There are plenty of good browsers for Linux (Mozilla and Opera) that can replace IE. Eventually, you can get users used to using Linux with KDE/GNOME and still give them the functionality they need.
One caveat: in my experience, leave the accountants alone. They tend to be moody, set in their ways, and can become quite a strain on your happiness if you try to mess with their routines.
The only thing I can offer you advice on are your custom apps. For those you can either just live with the fact that you need MS for them, try to find Open Source alternatives, or if written in-house, consider porting.
I hope this helps you a tiny bit. Best of luck!
you don't have to outrun the bear, just the slowest person in your group.
At this point you really don't .. because most of your employees also use Windows at home ...
.. as well as simple things like fonts not displaying properly .. websites not working properly (ie. streaming win media) (yes we know about crossover but BUYING a program to run FREE windows programs seems wrong) .. file system structure .. and believe it or not the 'ugly' interface was mentioned quite a few times. Yes GNOME looks nice - but its nowhere near as polished as Windows or Aqua ..
As an experiment some of our more computer literate employees switched to Linux - but in the end had to switch back due to a plethora of small but annoying problems.
These included opening files sent via email
I would call this a small company, but that actually makes the proposition easier.
First, evaluate the alternatives. What applications can be replaced with Linux equivalents and which can't. For the ones that can't, would it be cost-effective to consider limited licensing for those apps and running them from a Windows terminal service with Citrix Metaframe installed? Or would it be cheaper to by VMWare licenses for those users who absolutely have to run some esoteric Windows app.
When the alternatives have been considered, propose a pilot targetting a limited group of users to see whether the can continue their normal work routine on the new platform.
These are all starting points. The tough one is what to do about apps that only exist on Windows and are critical to the job the person is doing.
-David
* As is generally the case, my opinions do not reflect those of my employer.
I'm slowly migrating to Linux at my company. So far it's going pretty well. We had to add an extra box to everyone's desk in the call center due to [Non-Disclosure Agreement], so I spent $1000 for some old P233MMX w/64MB RAM, slapped Linux on them, and the users went to town.
After a couple of months with Linux in production I have to say that it's going well. Help desk calls are waaaaay down and users are happy. I've said this before, but here goes: users don't remember the last time they rebooted the Linux box, but sure remember the last time they rebooted Windows.
We're replacing Windows boxes right now by attrition, but plan to replace them all by the end of the year - OS, not hardware.
Since Office runs on Linux now it's at least theoretically possible to standardize on Open Office or something internally and have a single box with a single MS Office license converting both incoming and outgoing attachments to and from MS office format. When an open-source conversion utility comes out, you'll probably be able to abandon MS Office altogether but keep compatability with others.
Be aware that Citrix ain't really that cheap a solution. You'll pay through the nose for licenses (application licenses for each connectd user, terminal server client access licenses for each connection, citrix licenses for each user, and connection licenses for each computer). You have to get licenses for client machines (unless they're Windows 2k or XP which have their own) that connect to a Citrix server, which defeats the financial purpose of replacing Windows with Linux.
"Your biggest problems will come when everyone and their dog wants to install their personal stuff (screensavers, wallpaper, P2P apps, etc...) on their new Linux machines, then get mad when you tell them it won't work."
Screensavers? Have you SEEN the latest full Gnome install? I could spend an hour browsing through the stock screen savers! As for wallpaper, that's cross platform. A jpeg is a jpeg is a jpeg. P2P apps would be a problem if it wasn't for wine and people like Frank, who've figured it out for you. Closed formats are really the only thing that Microsoft apps can read that Linux apps can't. Even that barrier is crumbling.
Here is a quote:
Still, in my opinion, the great majority of end-users is still too dumb and this will cause you greater troubles, at least with all the time you will waste trying to explain'em that No, You Can'T Have A Dancing Lady On Your Desktop, And Even If You Were In A Windows Environment You Just Cannot Install It And Later Complain Because You Got a Virus!
However, have fun.-- There are two kind of sysadmins: Paranoids and Losers. (adapted from D. Bach)
First of all lay down your architecture. In our shop we started by loading a huge compaq server with memory. We went to a local used pc company and bought a whole bunch of used p350 machines at 100 dollars a piece. We loaded redhat 7.2 on the big server along with open office, mozilla and some other productivity apps. We turned on GDM on the big server to dish out x displays. Next we built a kickstart install to do the workstations. The kickstart does a standard minimal linux load with x and at the end of the install modifys the inittab to query the server for a display. These machines do nothing but X display so we wanted to capture the free cycles to run computatational fluid dynamics applications on. So we add the mosix kernel at the end of the kick start and boot the now running workstation. Another way to accomplish this is by using LTSP but it is more trouble than it is worth in my opinion. It is far easier to just load linux on the local disk. We totally control the desktop on each machine right down to the application icons...The moral of the story is "we are happier than pigs in shit with the outcome" and I have never had to show a single user how to use the kde desktop.
Got Code?
Here are a few things that I've used to eliminate MS products on my networks:
:)
For word processing I like AbiWord. The 1.0.1 version has just been released and it works quite well. It doesn't have *every* feature of word, no sane program would, but it does have all the features needed for general word processing.
I don't have much use for spreadsheets, but Gnumeric and kspread have worked fine for me. People have also mentioned that OpenOffice/StarOffice has a good spreadsheet.
For reading email there's Evolution, Netscape or Mozilla Messenger, and various others.
For database, use MySQL or PostgreSQL as the backend instead of access. Use HTML as a frontend so you can access it from any system, even Windows.
For webmail, take a look at squirrelmail. There are many other imap/pop/mtas that you can choose from to create your mail server. Of course, you'll replace IIS with Apache
For viruses you'll need to use the Unix honor system. su to root, choose a file at random then delete it, then email everyone in your address book with similar instructions. To mimic the crashes you can try turning off the power when you're in the middle of something very important.
Well, you must have realistic goals like: Reduce software costs, provide a stable environment, reduce support costs etc...
If you want to use Linux as the OS, you may get some opposition from the Windows fans. So, migrate using a gradual approach.
After you do all these things... the stability and usability of the user and server software should be evident and the switch to Linux as the OS should be fairly painless as there are Linux equivalents to the programs they've been running on the old Windows installations. You may find you have made some Linux fans in the office too!
Codifex Maximus ~ In search of... a shorter sig.
It's a very thorough overview of all the major steps (technical, mental, emotional, you name it) that an office must pass through in order to successfully dump Microsoft. It'll be very helpful to your cause.
We have 45 users, most local but three remotes in other states. We recently took a look at Microsoft's pricing, calculated our costs over the next three years, and ended up converting to Lotus SmartSuite. If the current StarOffice had been ready in time, it would have been a strong candidate, too. Keys to the process:
- Bring the users into the decision, not only what software to use but why.
- 3rd party training for software other than MS Office is available, even if not listed in their course lists. Ask. Our local Productivity Point has personnel qualified to teach the Lotus software and has complete course materials, they just don't list it in their offerings because there's so little demand. They were delighted to teach the courses for our users, at very reasonable cost.
- Having a backup process leads to user comfort. We'll still have four people with MS Office on their PCs, mostly administrative assistants and a lead customer service person. Their primary purpose is to provide access to documents which don't get converted by the time we remove Office from everyone else's PCs and to convert documents from outside the company which for one reason or another won't convert to Lotus cleanly using the Lotus software. This is a real benefit to our users and we wouldn't have gotten their buy-in without being able to assure them that we weren't abandoning their old documents.
- Don't rush the process. We started out by giving our users 60 days to convert their old documents to Lotus. We'll end up giving them an additional 30 days on a case-by-case basis. We installed Lotus on user's PCs in addition to MS Office so they could get used to the new software gradually. Once a user has been to training we made it clear that all new documents were expected to be in Lotus format. They have both the incentive and the training to make the change and it is working out very well. On the other hand, there has to be a due date or nothing will get done!
Our users initially resisted changing and why not? Learning new software, even as simple a change as from MS Word to Lotus Word Pro, is intimidating to someone who views computers as a tool rather than a way of life. We overcame their resistance by putting the facts before them: the lifecycle cost of MS Office over the next three years vs. the lifecycle costs (including training!) of switching to Lotus, Corel, etc. The savings were really very dramatic, particularly for a company like us which tends to keep using old software for much longer than the vendor would really like. Since we're a pretty open company anyway and take pains to not only present financial information to everyone but teach them how to interpret it as well, this had an impact. When you put it like, "we can spend the money on MS Office software and upgrade desktop PCs every five years, or switch to an alternative and keep to our three-year cycle", everyone had the same answer. They *like* getting new PCs every three years. The admin assistants *like* using shiny new Thinkpads which they can take to meetings and access information or take minutes with wireless connections to the LAN, etc. All of the productivity and convenience improvements we've made over the past five years took capital to implement, capital which in no small part would have gone into simply maintaining the software they already had. They didn't want to do that.
Once the decision was made we immediately chose a dozen key users and sent them off to a special Lotus SmartSuite class we had developed with our local Productivity Point franchise. A combination of the Introductory and Intermediate classes, it assumed that everyone knew how to use a mouse, access pop-up menus, etc. and concentrated on the differences between MS Word and Lotus WordPro, Excel vs. 1-2-3, and Powerpoint vs. Harvard Graphics. The class lasted three days and we had a very enthusiastic (and very relieved!) team when they got back, confident that they knew what they were doing and how to do it. They've been evangelists to the rest of the organization to the point where people were competing to get into the next class! A month into the process, people were competing to see who could be more 'MS Office free' and new documents were all being generated using the Lotus software. That was a month before the deadline! Now when someone from outside the company sends us a Word document (non of which, by the way, we've had the least trouble converting to Lotus,) people grumble about how 'backward' other companies are. Those who have regular contact with MS Office users outside the company are now evangelizing them and rumor has it that at least a couple of our business partners, faced with the same costs we were facing, are considering the same move away from MS Office.
The key is not that the Lotus software is so good but rather that it is good enough and *much* less expensive (and a key to that is that we buy IBM notebooks and desktop PCs, mostly because of their terrific technical support, and SmartSuite comes free with them. A real savings, that!)
Difficulties in our project:
- Some MS Office documents do not convert very well to Lotus. Most Word and Excel documents do (in the case of Excel, usually needing at least some clean-up by hand,) and Powerpoint presentations don't convert nicely at all. Don't even think about converting MS Access to Approach (but then, think about it: do you really trust anything more complex than a grocery list to Access? If its really important, shouldn't it have a real database behind it?)
- Some people just don't like change. We've got a couple. Peer pressure works most of the time but we have one granite boulder who not only isn't going to change, is senior enough that it would take tippy-top management ordering him to before he would, and then he'd just sabotage it ("See? Told you so!") Not a real problem, though: he doesn't really use the computer for anything other than e-mail, anyway (and half the time he dictates replies to e-mail to his admin assistant.) His assistant is very happy with Lotus and buffers between him and the rest of the company, so it works out.
Good luck!
What about the option no one seems to consider? Stick with what you have right now. It works today...it will still work tomorrow. Get out of the "gotta have the latest" mindset.
For most of what business users do, using software that is a year or two or five old is just fine.
Whats the best way to manage multiple user accounts over mutlitple linux workstations?
NIS is the proven workhorse for this task and I've had good times with it. LDAP is much more elegant, versitile, and scalable but is a little harder to set up.
Would you want to have to run XDM on every machine though?
Yes. Most people are comfortable with a graphical login. Also, gdm/kdm/xdm will give the users a very polished system to work with - which is always good when you're trying to convince them to change and learn new things.
Of course, what you do with dedicated servers and the machine on your desk is your own business.
and would you want /home to be a NFS share? or other nextwork filesystem?
Sharing home directories with automounted NFS is the standard way to do this. It's solid and works well.
The real question for this is how often do users need to acess each other's files, and would it be convenient for them to be able to log into other people's machines and have their desktop appear in all it's glory? If the users need to read each other's files on a regular basis or if roaming is conveneint, then NFS-mounted home directories are a must.
I don't know about Konqueror, but I've never had any such problems w/ GNOME. If the user doesn't know what to do, they could just go back to the Windows way - reboot. However, I don't see that as an issue because most distributions ship code which works just fine. If you must run CVS versions of programs, then it's your own fault. If a vendor's software is having problems, complain to the vendor.
You don't need to explain anything about permissions to users. None at all. Not to use it.
If you are talking about adding/removing packages, Linux is _much_ easier.
It doesn't take 50 pages of doucmentation to run XCDRoast or whatever your vendor ships. My CD Burner autodetected just fine.
In fact, so did my graphics card. Then, when I changed it, it autodetected it again.
Have you used Linux since 1995?
Engineering and the Ultimate