Development environments: Microsoft vs. Open Source

As we saw yesterday the much vaunted Unix skills premium over Windows is pretty small - 15% or so in an overheated market and less than that elsewhere.

Notice, however, that this information pertains only to larger organizations: people who can pay their neighborhood pretend-a-geek $25 per hour to futz with their Windows machines are generally convinced that's a lot cheaper than paying a formal IT labor retailer $180 and up for insurable skills.

Underlying that belief is an assumption: that the services have comparable value, and while I doubt that, this is another area where the data needed to know for sure simply doesn't exist. For larger businesses, however, we have lots of numbers - and their consequences are mind boggling.

These come in two parts: an easy part, and a hard part - and bear in mind that that I'm looking at this from the perspective of a niche issue: the risks associated with choosing closed versus open source software. There are, in other words, lots of other issues here - but they're not part of my agenda with this particular Thursday series.

The easy part is, well easy - it's about the minimums needed to operate the IT architecture at all.

To support the typical 1,000 user enterprise on an all Wintel architecture you'll need an IT director, two junior managers, about a dozen user interface staff (aka help desk people), five people in data center operations, all plus a couple of juniors to manage licensing, compliance and ever-greening - and, bottom line, you'll find 2o FTEs an extremely conservative minimum just to keep the lights on.

Replace that architecture with a couple of Solaris machines, 920 or so Sun Rays, 60 Macs, and perhaps 20 PCs and you're going to be paying about ten people, four of whom you don't actually need every day but have to have to provide off-hour coverage and a backup ready to step in when other team members go on vacation - or get hit by a bus.

Double the size of that Windows infrastructure and your non management FTEs go up nearly linearly - double the number of Sun Rays on the Unix side and you won't be hiring anyone new - Sun runs 33,000 Sun Rays with fewer than 30 full time staff.

Notice that most of the Wintel architecture money goes into desktop support, not data center services - but switching those desktops to Terminal Server/Citrix clients doesn't let you significantly reduce your FTEs because that support requirement comes from the Windows-Client server architecture, not the physical location of the device. Thus Windows thin clients save you some money on support, lots of money on ever-greening, and let the company save some more money on power and air conditioning, but on net you're probably looking at no more than 20% -certainly not order of magnitude change.

The hard part is, well, much harder even if you want, as I do here, to only look at one consequence of it.

The basic staff configurations in the two scenarios above are fundamentally different. On the Windows side I've suggested just enough people to run the typical set-up with the typical "base-load" applications: communications services, office services, some financial package integrated with whatever the major business application is, and perhaps some web services.

On the Unix side, in contrast, I've suggested that you'll be hiring people who won't have much to do just to spread the burden on off-hour operations while protecting the business against key man syndrome, vacation effects, and the occasional hostile bus.

Now consider a new application decision in both contexts.

Go to unsupported open source on something that's likely to become mission critical and you pretty much commit to having your people develop some in-depth expertise with it - but go proprietary and you know that almost any one of your people can play the trained monkey on your end of a support call to deal with emergencies as they come up.

In other words, the less flex time you have in your IT staffing regime, the better contractually supported proprietary software starts to look - and, conversely, the more unallocated time and expertise your staff has, the better both supported and unsupported open source starts to look.

In the end this is, of course, all about 3AM risk - the more you empower your IT staff and thus the less you pre-commit their time to busy work, the more software risk you can take and thus the better open source looks.

Notice that this result is nominally OS independent: it's as true for Wintel users as Unix users - but the real bottom line is that with Unix you've got skilled people with the time available to do this and with Windows you don't - because with Windows base costs are high; the workload's heavy; skill ranges are narrow; and your use of the same software, on the same hardware, run by the same interchangeable people as the other guy, condemns you to compete on nickels and dimes.