Unlike businesses or consumers of IT, governments often have unique regulations designed to optimize price/performance in purchases. One frequently encountered provision is that purchases over $X must go to public tender spelling out requirements.
We have seen examples where governments tried to skirt their own regulations by dividing up purchases into smaller units, something usually less efficient, and wording tenders so that only a single supplier need apply. The latter was the case in the province of Quebec recently when a raft of computers were purchased that could only be supplied by M$’s “partners”. A FLOSS consultant took the government of Quebec to court and won costs but did not get the tender award reversed because the computers had already been installed. In that case the court ruled the deal had been made “in good faith”. That could be a joke as purchasing departments are supposed to know the rules… They are not supposed to assume stuff.
This is a complex issue. It could be approached differently depending on whether a purchase is for a new installation or a replacement of existing technology. There is an increased cost to change things in either case but in both cases you should consider future costs as well. FLOSS should be the default and have an advantage considering future costs, simply because licensing fees per unit multiplies by units multiplied by infinite upgrades is a huge amount of money to be spent. Locking that in is wrong.
As far as I know the regulations requiring openness in the purchasing process do not require examination of future costs but they should. That was one of the reasons Munich went ahead with migration from that other OS to GNU/Linux. While migrating to GNU/Linux was a similar cost to upgrading to the next step on the Wintel treadmill, the future costs should be much lower.
Let’s work an hypothetical example. Suppose it costs $1000 per unit to take the next step on Wintel’s treadmill and $2000 to convert systems to GNU/Linux, combined with buying newer equipment in both cases. If only the present upgrade is considered it’s no contest. However, if future upgrades do not require additional hardware with GNU/Linux and that other OS does, it’s a simple decision the other way. Even if future upgrades both require new equipment, GNU/Linux should win: $1000 X n ($500 for hardware and $500 for licences) will always be larger than $2000 + $500 X n (added migration cost plus hardware) if you make n (the number of upgrades) large enough, 4, in this case (non-free – free = $500n -$2000). Unless you believe the world ends tomorrow, GNU/Linux should win on costs. I am assuming labour and hardware cost is the same either way and I don’t believe that. GNU/Linux has always been easier to install/maintain in my experience. GNU/Linux gives good performance on lesser hardware, too. If the comparison is between that other OS on thick clients and GNU/Linux on thin clients, n, for break even is 1 or 0.
The reasons maintaining GNU/Linux systems is easier are several:
- Mo accounting involved in licensing. If you have the software, you have the licence.
- Far less malware, likely 1000 times less.
- Better performance in beta and production. Installation/upgrading software is just copying. For large roll-outs to which this post applies, disc imaging is used to the time to copy is less relevant. Both systems will take about the same time to roll out except that that other OS just uses a lot more disc space to get less software on the system.
- Fewer re-re-reboots which waste time.
- Easier control of multiple systems in real time. No need for reboot to propagate changes in many cases.
- Package manager for the system as well as applications.
I believe we are at a stage where governments around the world are going to put aside FUD and look at the facts in choosing/purchasing IT. Any OS can function. GNU/Linux costs less to do the job. The FUD that no applications are available for certain specific tasks is nonsense. Governments are larger than the corporations producing non-free software so they can produce their own software at much lower cost especially if it is shared amongst governments.