I know a dozen or more people who work as Unix sysadmins for big organizations in which most of the really critical stuff runs on Unix, but Windows get pretty nearly all of the money, all the people, and all the attention. In every case the Unix gear is either reaching or already well past its reasonable end of life, and IT management's multi-year response to the issue has been to promise to convert everything as quickly as possible to Windows servers - while achieving nothing.
In about half these cases, furthermore, the current Unix hardware was purchased during the 2001-3 period when late nineties efforts to convert the software first to NT and then to 2000 failed and something, typically disk or processor limitations on the early to mid nineties Unix gear used for the key applications, triggered emergency hardware upgrades.
I know one such company whose development efforts came under the sway of a global outsourcer in early 2000, reacted to an EMC failure in 2001 by leasing a new RAID array from HP, brought in a pair of IBM 680 machines in the 2002/3 budget year to replace its K-580s, but then failed to get the software migrated for another year. Today they're on their fifth consecutive lease extension on the Power4 gear, still running AIX 5, still paying a former colleague there a per diem to admin the boxes, and still committed to moving their ERP/SCM databases and BI applications to Windows servers.
He's more than paid for his retirement out of this, but if you're an ordinary employee in a similar position, what can you do?
I think the short answer is quit, and go find a job where your abilities will be valued - but the longer and harder answer is that while it may not be possible to turn the psychology of this around, it may be possible to use the system to get yourself some new gear - so at least you can improve performance, learn something new, and reduce your stress levels while doing it.
The cost of new gear keeps coming down - right now, for example, those two P680s could be replaced by a couple of T2000s backed by a "thumper" ZFS appliance for little more than three months in maintenance and support costs. Management won't do that, however, because the maintenance budget is cast in stone, fronting a proposal for new Unix gear requires them to question the value of their effort to convert the applications to Windows, and "everybody knows" that Unix costs too much anyway.
Those 680s, for example, cost the company a couple of million bucks to put in and get running -while an ad in this morning's drug store flyer offered a PC for $195, and since they're both computers the only difference is that one runs Windows, right?
Worse, nobody pays attention to stuff that works. The incentives to get you some new gear and opportunities just aren't there: you don't get face time with senior management without a crisis, and your stuff may do almost all of the heavy lifting on the revenue side of the business, but if it works well your own bosses in IT may barely know you exist.
To beat this combination what you need is a one-two punch: a crisis with an easy solution that just happens to give you what you want: some shiny new hardware, Solaris 10, ZFS; possibly even someone to share the load and a new software challenge or two.
The trick here isn't in creating the crisis - you actually live in a perpetual state of crisis, even if your bosses don't know it. Response is never fast enough for users, backups are never secure enough - there's always somebody who wants something more. No, the trick is to leverage the used equipment market to unlock the budget and thus allow management to get you what you need.
The market for big used gear is seriously depressed by the speed at which the cost for small new gear has come down. In most cases, for example, you can now get a brand new quad Xeon Windows or Linux server with serious memory and disk resources for the price of a new disk pack for your production gear - and I just saw an ad offering a Sun 6800 with 24 CPUS and 96GB of RAM for $12,000.
Meanwhile new mid range gear has come down even faster - and both trends are about to accelerate as Sun releases the UltraSPARC T2 to production.
So here's how take advantage of this: get some cheap used gear in as an expedient, short term, solution to a current processing or storage crunch, and then leverage its maintenance cost to get the new gear you need.
This works because the used gear meets expectations about Unix cost: you are buying a million dollar machine - for perhaps $20K. Thus instead of seeing your idea as a challenge to their beliefs, your bosses will see the proposed acquisition as a short term, low cost, low risk, way of showing user management that they care: getting the primary applications new servers, upgrading to the latest RDBMS releases, improving security, and so on - a million dollar validation for a few thousand bucks.
What happens next is that the trap closes: because as soon as you get the new old stuff ready to go into production, they're going to have to sign the hardware support contract - and because that's new money coming out of their budget, they'll listen to the proposition that buying brand new gear with three years of warranty support for production use while keeping the new old stuff off support, but ready as a backup, makes sense.
The numbers will work: the company gets savings; you get brand new gear, warranty support, a quality backup system, and a chance to see your old HP or IBM gear in the back of a truck - but, believe it or not, it gets even better.
How? well, with budget sanctity now broken, the monies they've been spending on supporting that older IBM or HP gear is now up for grabs - meaning there's an opportunity for you to sell them something else they're going to want around this point in the process: a backstop for you. Someone you can see as a colleague and they can see growing their staff while strengthening their control over your work.