Unix: An out sourcing vaccine?

- by Paul Murphy, -

One of the unfortunate realities of the Wintel monoculture is that expectations are set mainly by ads, the Sunday Supplements, and stereotypes like Marshall, the Alias super tech, but delivery is constrained by reality. As a result the MCSE has no chance of meeting user expectations and must always operate in a kind of rear guard mode balanced between short term fixes and Microsoft's promises for the next upgrade.

This inability to meet expectations has many negative consequences, among them an increase in senior management's vulnerability to an out-sourcing pitch. What happens is that general dissatisfaction arising from unfulfilled expectations greases the way for acceptance of an out sourcing pitch that is actually based more on appealing to behaviours learnt in high school than business fact.

The fundamental emotional appeal is that senior management can distance itself from the people who run IT by passing the burden of association with IT to social peers working for the out source contractor. Since this reality is hardly an acceptable business rationale, two levels of rationalisation: cost savings and service equivalance, are generally applied to cover it.

Of these two, it's service equivelance that's truly important and cost savings that's most talked about. The underlying issue with service equivalence is IT commoditization. Fundamentally, if your company has the same Microsoft software, the same hardware, and the same MCSE skills as everyone else then there isn't any competitive advantage in IT and cutting its cost by contracting a specialist service provider makes perfect sense --especially if that out sourcer also promises to do better than your own people.

That's why out-sourcers almost always promise to provide highly reliable services using the very latest in Wintel gear while cutting costs. In practice, I've never met an executive naive enough to believe such obvious nonsense, after all IT commoditization means first that the out sourcer's costs can't be all that different and secondly that the out sourcer's skills can't be all that different either. It takes, in other words, extraordinary incompetence on the part of the internal IT group for an out sourcer to be able to add a layer of management and still improve service while reducing costs.

Unfortunately rationality doesn't stop management from signing such contracts because they're not really doing it for business reasons, they're doing it for social reasons: IT, after all, is nerd central, something to be avoided at all costs. Thus senior managers seldom believe either the cost or the performance claims on offer. Instead, they invoke cost savings as a business rationale while telling each other that their decision won't affect IT costs or performance because IT will always cost more than expected and deliver less, no matter who runs it.

That's totally wrong, of course; and the attitude itself signals a disconnect between top management and the flow of information within the organization, but marching on up to the executive suite and telling them to do their job isn't going to help you keep yours. So what can you do? Well, if it's not already a done deal and you have at least six months to act, then the right answer may be to put yourself at risk through a combination of user education, Linux, and decentralisation.

I know all three are anathemas to most MCSEs, but you need to understand that the more you centralise control, the tighter you lock down those user PCs, and the closer you get to bog standard Wintel deployment practices, the easier the whole thing will be to out source and the more reasonable that out sourcing will look.

To head this off what you want to do is break IT commoditization since that's the basis for the current "utility computing" rationale. Now, in fact when Scott McNealy first used that term it meant something about reliability and having IT services there when you need them, but it's being used as a marketing slogan covering the abdication of IT responsibility to a third party or "IT utility" and that's what you want to stop.

To see how, consider the lessons learnt from nineties style organisational IT out-sourcing; every single one of those deal announcements should have been seen as a signal to sell that company short because willingness to out source IT almost always turned out to mean that management didn't value its most significant organisational assets: people and information. In most cases implementation of those contracts led to three to five years of nominal stability in the formal IT budget expense even while the contracts grew more holes than a swiss cheese as exceptions, special authorisations, addendums, and one ofs were tacked on. In most cases, however, actual IT operations started to diverge from the contract within minutes of the internal announcement as line management -the guys who actually make the money- set out to pretty much duplicate everything they liked about IT before, but this time without central co-ordination, support, or budget authorisations.

Two of the best things about these subrosa IT activities are first that they tend to be cheap and second that they tend to be effective. You don't get a line manager investing in a couple of BASIC developers for SQL-Server on Windows Advanced Server if he can get someone to bang the thing together using Java with Apache and MySQL on Linux. IT may care about standards, but a line manager's attitude is usually much simpler: he just wants his stuff to work, he wants it now, and he wants it cheap.

That attitude makes him the guy you want to work with before you get out-sourced. Build those bridges strong enough and these guys will get so heavily invested in IT that they'll stop any out sourcing proposal long before it becomes a threat to you. In effect what you need to do is break senior management's IT commoditization rationale for disassociating themselves from IT by making IT strategic to middle and lower management.

So how do you do that? More of the same Wintel stuff is just going to make things worse. You have to break out of the mindset with user education, Unix, and decentralisation. Education so they know what can be done, Unix so you can do it, and decentralization to make it theirs. Do what you can, whatever organisational level you work at, to help users make effective use of IT. Learn how, and when, open source tools like Apache or Linux simply make the most sense and then go tell everybody about it; volunteer to work with the nastiest, most demanding, most anti-IT, user managers and be prepared to fight your own bosses to deliver what the users want.

That's risky, but remember that user management doesn't care about commodity IT services any more than senior management does; from their perspective the CIO is a contracts administrator who works for the enemy, standards are things that raise costs, and vendor viability reviews are excuses to waste time. They do care, however, about getting things done, so if you step outside the Wintel box to make things happen they'll see a benefit - and stop any out-sourcing effort in its tracks.


Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 20-year veteran of the IT consulting industry.