% fortune -ae paul murphy

The missing demonstration project

Last week frequent contributor ross44 raised the issue of the missing Sun Ray demonstration project:

So - it seems that once again, this time in the public admin sphere, we need a reference installation. Where is that (US) state (or Canadian province) that is saving immense sums by using [insert favorite nix] on [delete one:] SunRays / cheap PC's?

He's right: you can't conclusively demonstrate the risks and consequences of a non Microsoft decision if you can't point at some large and successful users of the alternative you want to suggest.

In the specific context of Sun Rays there are two reasons you can't use existing larger installations as technology demonstrations. The obvious one is that the people who use Sun Ray usually do so for its security advantages - and are correspondingly unlikely to want to participate in a highly public demonstration project.

The unobvious one is that the technology works well, and for a long time - because this means that well established systems of the kind you'd want to point at look oddly archaic to the uninformed. Sun would not, for example, want to point a bunch of superficial, and generally wintel biased, reporters and commentators at a large school district successfully running Sun Rays since 1997 because the servers are getting old, the original 14" Sun Rays are still in use, district management expresses its contempt for IT by never thinking about it, and most of the school principals know their janitors but not the IT guy keeping everything going.

You'd think, however, that the obvious advantages would mean that more large commercial or government organizations would be adopting it every day - so why don't we see that?

In its generic form the problem I think is most relevant is that a decision maker who goes against popular wisdom is immediately considered personally responsible for any negative consequences, real or imagined; is guaranteed to become the target for numerous ad hominem attacks; knows that the decision will be reversed ten seconds after he resigns; and can't expect any credit for the decision when it turns out to have been right.

In the special case of a choice between Microsoft's client-server architecture and the Unix model combining centralized computing with decentralized control there are additional gotcha's to worry about. In business, as elsewhere, squeaky wheels get the grease - maintain the continuous low level failure associated with the Microsoft architecture and you'll get lots of face time with senior management along with more money, more staff, and ever increasing control over business processes and users. Bottom line: in IT failure succeeds and success fails; do your job well: deploy Unix, centralize computing while getting your sysadmins working directly for users, and what you'll achieve is your own invisibility - no squeak means no grease.

With that in mind imagine yourself facing a decision between getting the job done right, but at significant cost to your own career prospects, or going along with the crowd to spend somebody else's money growing your career - and incidently having fun playing the technical hero to upper management and selected consultants and vendor staff.

It's easy to decide to do the right thing - after your morning coffee and in the presence of friends with less on the line, but another thing entirely when you're laying awake at three in the morning worrying about the tuition bills you're going to face when the kids hit University.

Add the fact that longer term group decisions tend to go those advocated the weakest positions and you can see why a Sun rep brave enough, and foolish enough, to tell the Illinois central management people that Sun Rays could save their taxpayers a couple of hundred million over any five year term would have been considered insane - and given the bums rush out of town rather than a fair hearing.

The bottom line is that money saved on IT is money that doesn't go into the IT budget - and therefore that saving money isn't in the IT director's personal best interests. Thus what really drives the absence of large scale commercial Sun Ray success stories is simply that the Microsoft mainframe tarball is in nobody's interest except that of the IT people making the decision, while the Unix model is in nobody's interest except for user's and shareholders -and they don't get to vote.


Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.