Researching changing organisation from Win to Linux

(Mods, please move this to the other operating systems forum if it more appropriate in there, but since this covers more than just Linux I put it in here)

Ok guys, I've been tasked with researching the possibility of changing our departments standard desktop image from Windows XP to some variant of Linux.
The driving force behind this is to lower our software licensing & support costs, so were quite willing to pay for the right stuff, as long as the sums work out in the end.

I'm researching a couple of things currently, and will do some more detailed work if this starts to get serious.

1 - Replace Windows XP with a Linux distro.
The boss has specified that the closer this looks like and works like windows the better, as there will be a lot of pressure from other business units not to change, as they are comfortable with XP.
I personally think that selling the move from Office to Open Office would be the harder issue.

2 - Possibly replace ZCM with an OS product (if one exists).
Currently I build our SDOE images and were currently in the process of reimaging all machines in the dept via ZCM.
While ZCM does the job pretty well, we are interested in looking at other (cheaper) alternatives if they exist.
Whatever way we go, we needs to be able to push out application bundles to designated users / usergroups as well.



Currently the theory is to research and replace as many apps as possible with OS equivalents, but there will be a number of apps that we wont be able to do this for.
The current suggestion is that these ones would probably end up running in Xenapps (Citrix) sessions, although I personally think that we will still have to run a number of XP boxes for some specialised applications.


So......
I have only been very brief as its early days yet, but I'm interested in any opinions as to what software to investigate, where to look, who to speak with, etc, and I'm espically keen on hearing of any experiences good or bad if you gone this route.

 

Windows XP isn't super expensive, Have you thought about keeping it, and just replacing what apps you can with OSS alternatives? Such as OpenOffice, GIMP, etc.

Have you got much Microsoft server infrastruture? AD/Exchange etc ?

Ubuntu 9.04 is pretty easy to connect to AD these days, and SLED ( Suse Desktop ) is supposed to be easier again.

If you have a lot of legacy Windows apps to run, then any savings you make on a $200 OEM OS per seat, you might lose in configuration/hassles.

 

Agreed, I also suggest keeping the host OS for now and migrating to OSS that is the same between the various platforms. Once most of your basic office functionality is on Open Office etc, you can then start to migrate to a different host OS. Baby steps, baby steps

It's funny how set people are in a particular host, when in reality they only need to be able to load Word/OO and send email etc.

 

What's the business driver to move away from Windows?

 

Run a pilot group. Run up say 3 linux machines to have exactly the same functionality as their equivalent XP machines, give them to one person from different department's for a week (make sure you pick a receptionist too) see how long it survives

That's generally enough to keep people on the dark side

Cue elvis in 3...2...1

 

I really don't want to get into the whole TCO argument because it feels like beating a dead horse. But you need to consider non technical things like (but definitely not limited to); availability and support of of your industries software on the platform, finding users proficient in the linux equivalent's and/or training new and existing users to use the new systems.

Don't forget your sys admins, they also need "non standard" knowledge bases and qualifications.

Also, keep in mind that building a system and maintaining a system are two different skill sets. You might want to get a company in to design (and possibly assist you with support of) your new linux framework. This will save you countless hours of research and procedure writing if you're not already up to speed.

 

Spelled out in first post.

If the OP is comfortable divulging the information, what industry are you in, and roughly how many employees?

 

Saw that, but in my mind that's not a business driver, that's an IT driver - the end-users don't care about the cost of the software but they will care about the change that (in their mind) is "imposed" on them.

 

How do you know that the business has not come to him requesting to get overall expenditure down, and this is one of the solutions put forth?

Cost and budgets are very frequently determined by the business. And particularly for small-to-mediums who don't have much leverage with vendors, switching core components of the business free and open source software can be a way to funnel dollars into more critical parts of the business.

I've seen it happen many times before (and assisted as a consultant many times) in exactly this sort of thing. And in my own personal experience, cost has been either a large or the primary driver as defined by the business (and again, far more common in small-to-medium businesses, where changes to annual budgets are felt very quickly by the whole business).

 

Considering you've got ZCM (i'm assuming a Novell shop here) and you're wanting to run linux, sled11 would probably be a good starting point to look at and depending on your Novell licensing you might be entitled to it already

Also you might want to have a look at Zen application virtualisation, you can build apps on a clean XP machine and in some instances they will work under wine

If you're running a Novell backend, bear in mind that the seamless login and other niceties users are used to don't work quite that nicely in Linux, not to mention the Novell linux client is only available for SLES (I havn't tested it on anything else though, but that's what it says on the site)

From my own experience though with trying Linux out in corporate office environments, as much as I love it and want to tout the Novell stuff, I don't think it's quite as streamlined and nice from an end user point as XP, openoffice has some really annoying quirks as well

It may cost you more in the long run in support and issue resolution than your xp and other licensing, not to mention that Citrix licensing is a bit frightful

 

Citrix licensing on Windows is frightful - mostly because you're not just paying the per-user Citrix cost (last I checked we were looking at ~$250/user in Citrix for < 100 users). You're paying Windows CAL's, TS CAL's, special Office CAL's (now with 2007) and running a Windows TS farm whilst you're at it.

Citrix is pretty awesome, but christ is it expensive.

 

Thanks for all the replies, most helpful.
I have read everything and taken it on board, even though I haven't responded to it all.

I think the idea is that if we go to the expense of changing, were better going the whole way.
In for a penny, in for a pound...

Yes, we have a shitload of MS servers (and Linux and Unix) and we use exchange as well.

Thats a given.
It would be quite a few more than 3 machines.

Thats good to know.
Ive now comfortable to ZCM, so one less thing to get to grips with, the better.

Thanks, will check this out.

Yes, whats perceived as the easiest groups/sites would be done first - if we go ahead.

Personally, I think selling this to the users will be one of the hardest things...

Thanks for that tip.

Were actually in the middle of rolling out Office 2K7 at the moment.

I'm under no illusions that this will be a big expense.
Something that will be decided above my pay grade though.

It will be a big learning curve for most of us Sys Admins, not to mention Desktops and Helpdesk as well.

We have a history of brining in outside help for getting specialised major projects started..
It does save in the long run.

Lets just leave it at around 2000 users spread over a few major and a couple of dozen minor sites.

I think we need to differentiate between the end users and their managers here.

I really don't know who (beyond my manager) is pushing this.

Thanks, will login and check that out in the morning.

Interesting. Thanks.

We have quite a bit on Citrix at present anyway, with a fair bit of spare capacity. I doubt if our Citrix licensing costs will change much either way.

The closer it is in operation to what we have now (from a users perspective) the better.

Yes and no.
The users opinions are important, but if we left them to make their own choices unaided...

Were already heavily into Citrix, so I cant see that changing unless there are big savings to be made.


So keep it coming guys, its been quite informative so far.

 

In my direct experience small businesses are the easiest to do "top to toe" Linux conversions on. Mediums are just about a no-hope (most are at the total mercy of the greater market and their vendors/support). Large businesses are a 50/50, depending entirely on three things:

1) Their type of workflow
2) The intelligence and motivation of the IT staff to help "at any cost" to make it happen (which includes both technical and non-technical/political/emotional issues)
3) Backing from management

And in my not so humble opinion, you need to work on those in reverse order. Starting such an adventure without the support and understanding of the top dogs will get you nowhere. Likewise, having IT teams full of point and click monkeys with no desire to expand their skills outwards (or worse, insisting that their current skills out to translate to a different system seamlessly), then you'll get nowhere.

Try and sort those two out first. Once they are concrete, you've got an almost guaranteed success.

 

Well I'm optimistically confident that this isn't entirely a flash in the pan idea, good show

This is the trick. How do you get the top dogs to know or care about such an adventurous project?

Bottom line it with cold hard cash. Show how it will be less expensive over 1-5 years (don't forget to add in typical microsoft new version inflation, ie more cost for less features so higher overall cost eh exchange 2007 no longer coming with outlook licenses requiring you to purchase office AND a cal per user).

Also tell them they can keep their vista viaos

 

Agreed. And secondarily, be prepared to talk management BS. If they need "forward thinking, next-generation green mobility and upwards acting synergy", then show them how this will achieve it.

Find the BUSINESS goals (of which budgets may be one), and show them how your strategy aligns with them, and enables them. You're going to need to talk a lot of fluff and dollars before you get techy.

 

I guess the biggest issue is if the OSS applications are suited to the task at hand.

One advantage of going to OSS applications first rather than completely changing platform is that if the rollout is unsuccessful (issues are found with the applications), it is easier to revert to the way things were.

However, rather than just 'Linux will save you money because you don't have to pay licensing costs', there needs to be other benefits of using Linux identified (perhaps an ability to do a particular task much easier than with the current setup). Basically, what can't you do now that you can do with Linux that would improve your business operations?

Also, you may need to identify any concerns that might exist (familiarity of desktops, suitability of applications, whether you are locked into one provider for support, etc).

 

I used to work for a publishing firm (2004 to 2008) that relied on Windows 98SE and Windows XP machines with MS-Office 2000 from 2001 and 2007.

In mid-2007 due to budgetary reasons the company decided to switch to a new operating system and Open Office. My friend who was the I.T. head at the time was a BSD fanatic, but he wanted Linux for the employees because he felt that BSD was not what was required for a desktop O.S. So I suggested to him to use Ubuntu (been dabbling with linux since 2005), which he gladly adopted (6.06.1 LTS) due to the readily available online documentation and the supposedly huge user base, which for him translates to less technical issues, since he was confident that googling for solutions or searching the ubuntu forums were the only things he needed to do to solve problems. (Note: Those were his reasons not mine. I suggested Ubuntu because of what I read on the net, but in hindsight we did also try Debian and SLED and even Fedora (Red Hat connection, perhaps?)).

As for "selling" the switch to my fellow employees..he adopted a sink or swim approach. Yeah I know it seems very cruel, but long story short the employees adopted to it with as little grumbling as possible. Of course it did help that we told them in advance what O.S. we were adopting and we had demo machines set-up for them to tinker in their free time.

 

This is good advice, and highly recommended. Having an "office expert" per location is mandatory for any software rollout.

The single thing that turns people off any software (open source or otherwise) is not having someone to turn to for instant-gratification-assistance. If people can't get help to get by the little things, they'll abandon new software in a heart beat.

Choosing these people can be tricky if you don't work with them a lot. You'll need to find someone who is not only technically competent, but also patient. (Every office has that one staff member who is always "too busy" to do anything. Even if they're a certified genius, don't make them your local software guru).

But I've seen software rollout projects that were initially very tricky succeed with flying colours just because this sort of approach was taken. Find these people, train them up (and make yourself available to them at any cost early on in the project), and your job becomes a whole lot easier.

 

I'm tired of this standard response, what support? You buy a support contract from a VAR, IBM, Red Hat, Novell or any number of other companies moving into that space if you require such support.

Any sys admin worthy of employment will have some *nix skills, as for qualifications MS<whatever they are called today> qualifications are ignored in preference of experience the same as any Red Hat/Other Linux certification is.

Knowledge Bases, is dependent on the distro of choice used, but all the main distros have very good knowledge bases, the more business orientated distros are on par if not better then any MSDN Technet Offering I've seen or used.

 

What support? I said it right there, your industries software. Are you going to tell the graphics artist to ask Adobe to support them running photoshop under WINE? How about Autodesk for 3ds max under linux? Or xyz merchant to get their point of sales program rewritten to run under linux? Is Red Hat going to do that for them? How about those 'linux equivalents'. Ever met a professional artist who works on serious projects who uses GIMP? How about some linux drivers from wacom for their Intuos4 tablets?

Come visit the real world sometime. "Some unix skills" is not what you want when one of your production servers has a catastrophic failure of some package or other. You don't have a couple of days to mess about and google solutions, post on mailing lists or wait for support in America to wake up.

I actually meant "knowledge base" as a persons experience and knowledge of a particular system. eg Every decent windows admin I've met has a personal "knowledge base" far exceeding that of what you'd find in one place on the web. I wasn't talking about some wiki. I'm saying you can't as readily find people with the skills to administer linux servers who are also willing to accept the pay of the SME space.