-   Linux - General (
-   -   Representing Linux... (

rich_c 07-04-2012 02:06 AM

Representing Linux...
I'm due to meet with some mid/senior managers next week to make a case for migrating to Linux desktops in my department. The main thrust of my case will be efficiency gains in that we are a department that works a 24/7 shift pattern. With the legacy desktops we use at the moment, the machines have to be rebooted at every shift handover. This wastes typically 20 minutes per person. The cost of this over a month/year is enough to get some attention, I feel, so my 'argument' is that using Linux desktops will reduce or eliminate this reboot/login requirement in that the PCs can be left running for days/weeks with multiple users left logged in ready to ctrl alt f7/f8 etc. to their session. Requirements for Windows specific software will be met by using existing virtual desktops which we access via RDP. Also to be mentioned is the cost saving to be had by delaying the need for a hardware refresh.

I'd be grateful for any tips on how to approach this meeting. (At the moment, I intend to demonstrate some Linux desktop capabilities by using my personal netbook...) Also, is there any other aspects that managers of this level are likely to be interested in hearing about? TIA.

TobiSGD 07-04-2012 02:18 AM

If there is also a change in software (besides the Windows VMs) then they may ask you about training costs. Also, have a look at training costs for the admins and may be (if you plan to use Red Hat) the subscription costs.
I would think that the main concerns will be the overall costs and savings and how long the downtimes and training times (times with decreased efficiency in the workflow) are.

rich_c 07-04-2012 03:01 AM

Good points to raise. As I see it, training will be a case of learning the system as we go. Initially, our usage will pretty much be logging in, starting an RDP client and accessing our virtual desktops.

As far as admin is concerned, I don't know if they'd consider it as viable but for our department I feel we have enough people with passable knowledge of desktop Linux to manage the small number of machines ourselves. Our Unix/Linux admins have enough on their plates at the moment without being asked to look after our desktops so I don't think asking them is an option.

Subscription costs is an area I've not yet got any info on. I've initially approached Canonical for a bit of advise on getting this idea off the ground. (Chosen mainly because most of the above mentioned Linux guys in our department, including me, are Debian or derivative thereof experienced.) I don't suppose they'd be interested in providing cost estimates until everyone has got a clearer idea of what will happen in practice...

Cultist 07-04-2012 07:45 AM

You might also want to be ready with a list of the FOSS to replace Windows-only things that your execs will want, like Libreoffice instead of MS Office. And is there any specialized software that your workstations run that doesn't have a Linux alternative?

And are the servers at your workplace running Linux or Windows? This could present another set of problems, depending on the answer and what the servers handle.

pixellany 07-04-2012 09:57 AM

gut reaction (imagining myself running that department): I could care less about reboot times---in any OS, the reboot time is insignificant compared to the whole workday. I want to hear about COST and PRODUCTIVITY.

Cost: You say the SW is free, but what is the labor cost to retrain everyone, maintain the systems, etc.?

Compatibility: You say that Linux will be compatible with MSOffice files--I have read that this is not 100% true.

Keep adding to this list---for everything that YOU think is a benefit, think of what the counter-argument may be.

rich_c 07-04-2012 10:20 AM

As far as servers at work go, it's a mixture of everything from IBM mainframes to windows servers with Unix and Linux in between.

The reboot time save thing has more to it than 'just' the cost in terms of time lost. It's also significant in that if an incident happens while we're rebooting, it could potentially cost a LOT in terms of fines due to missed SLAs etc. caused by us not being able to rectify or report the situation properly...

NyteOwl 07-04-2012 01:15 PM

Most relevant points have been mentioned above so I'll just add this: your desktops take 20 minutes to reboot? There is something seriously wrong with your hardware or software. Either that or you load and open enough software to run NASA at start-up :) Sounds like you might seriously benefit from new hardware or at least SSD's for boot drives.

As for incidents during rebooting, if you mean individual machines problems then that should be resolved with backup images (which I assume you keep). If you mean the organization being effectively down for the shift for 20 minutes, then that can be avoided by a staged shutdown/reboot so there are always some machiens available during the changeover of shift.

Not sure whay you'de need a reboot though. If it's just to track employee usage just logout/login routiens sohuld do.

salasi 07-05-2012 01:18 PM


Originally Posted by rich_c (Post 4718848)
As I see it, training will be a case of learning the system as we go.

In the medium to long term, this is not an adequate description. When people leave your department and have to be replaced it is easy enough to get people with Windows experience (nah, it actually doesn't make it significantly more difficult than just getting people, except in certain specialised areas in which you can replace 'Windows' with 'Mac'...these days the question 'Can you use a computer?' is a bit like 'Can you use a telephone?' would have been years ago, but confront the people who claim that they know all about every kind of computer from Baby onwards with something on which the wallpaper has been changed and they'll fall to pieces....sorry for the slight exaggeration!).

So, you'll have to deal with the question 'What do we do when we recruit and don't find people with...what did you say it was called...Linux experience?'.

If the answer is an in-house manual, then the follow up will be '...and how long will that take to put together?'

If they are smart (I'm only going to place a bet if you give me odds) they'll ask about:
  • support (ie, even if we don't want to do this, would we have the option of paying for support?)
  • support horizon; if we install 'distro X' today, how long before we go through a new install cycle because there is 'New Stuff' (TM) that has come out and our 'Old Stuff' (not such a TM) is no longer supportable

and maybe even malware and the like (Linux FTW!)


Compatibility: You say that Linux will be compatible with MSOffice files--I have read that this is not 100% true.
Even MSOffice isn't 100% compatible with MSOffice.
  • obviously, there is the issue of different versions. Well, obviously MS couldn't pass by a chance like that to presure people into upgrading
  • and then there is all the sort of printer and paper size dependency stuff that stops the MSOffice files being a sensible way of interchanging data that you want to print out the same way on more than one machine
  • and fonts...if you use fonts that aren't part of the default install for your version of Office, the document can fall all over the page on a machine that has different fonts installed

So, yes, this is a problem, but it is a problem that you may already have to a greater or lesser extent, depending on how much you exchange documents in and out of your organisation and the extent to which these are 'presentation-class' documents.

rich_c 07-13-2012 10:44 AM

Well, that went remarkably well! There is a very strong likelihood that we'll have pilot machines up and running within weeks... I got the impression that the management here were already well on the way to working toward Linux desktops anyway & all I provided was a nudge for them to go "Yep, let's try it." Based on that, if anyone wants to have a look at the presentation I put together, just PM me. I would publish it more widely but stuff produced on work time for work use is a bit of a grey area in terms of 'ownership'.

guyonearth 07-13-2012 12:36 PM

It shouldn't take 20 minutes to reboot a system, and there should be no compelling reason to reboot Windows machines every day. Your machines are either mis-configured, outdated, or overloaded, and switching to Linux may not solve your problem. In a more general vein, mass migrations from one environment to another are always potentially fraught with problems, usually in replacing or implementing something specific to the existing system. This is something that should be approached in stages, with a lot of testing first. There have been a number of migration boondoggles that have happened when large entities decided to go from an all-Windows environment to an all-Linux one. Sometimes it just doesn't work. You'd need to do a careful inventory of what HAS to run on the existing setup, and whether EVERYTHING it does can be implemented on Linux desktops reliably. There's really more to it than just demoing OpenOffice on a netbook, (not that that's all you'd do), but I've seen similar sales pitches fail because the Linux advocate really has no idea what actually needs to be accommodated, or they have no answer for how to implement specialized Windows-only applications either natively or in a VM environment.

jefro 07-13-2012 10:35 PM

I never got the reason to reboot at every shift. Why are you doing that? That should have never started. Superstition?

I have worked in 24/7 manufacturing for decades and have never had to do such a task on any OS.

There is some advantages to thin client but that doesn't affect this reboot issue.

rich_c 07-14-2012 10:00 AM

In short, the reboot thing is to pre-empt our machines becoming unusable at unexpected times during the day. This has been experienced time and time again if we fail to reboot at the start of every shift. This is on XP machines but we also have Win7 machines that we have to regularly reboot.

As for the attention to detail that would go into a pilot migration, I have no doubts that it will be done properly and with due consideration/testing. As I mentioned, I got the strong impression that some thought had gone into the likelihood of this happening at some point even before I raised my suggestion.

jefro 07-14-2012 11:44 AM

I doubt linux will solve this issue. We used to be a major user of QNX and it is almost bullet proof. I don't know what you are running but if you are using windows to control some machine, it won't be easy to convert. Many of the systems we have went from qnx to a java app on windows. Talk about junk but we needed some features.

I still don't get what the problem is. We have maybe a million xp systems and they run everyday without a reboot.

rich_c 07-14-2012 01:58 PM

Well, the people I have spoken to are willing to give Linux more than a fair go. I certainly got the vibe that all else being equal, they'd prefer a solution that would mean they don't have to deal with Microsoft. Of course they are realistic in that they appreciate that migrating to another solution will take years but I was quite surprised by the willingness to start down that road.

All times are GMT -5. The time now is 11:47 PM.