A day in the life

It's 5:28 AM, Thursday, August 13th. I see my favorite cartoonist has made his annual fund raising target. Great!

There's an email reminding me that I owe some work on a public policy web site I drafted some time ago; another from an editor complaining that a masterpiece of mine is too long and has too many compound sentences (well, duh!); one from a friend announcing that his firm has been selected to bid on building part of the latest provincial health care system - not that they have a chance: the same people who were in charge of the last half dozen insanely successful fully integrated health care systems projects are in charge of this one, and they only buy from each other - and one asking if I know where the writer can get realistic pricing information on a p6 595.

Here's my response to the latter - and if anyone has better information I'd be grateful if you'd send it along. Please:

No idea - sorry.

IBM doesn't normally publish pricing on these and there aren't enough of them to establish a real "street" price.

There is a June 2008 TPC report (costs dated May 2008, 64 core, 5Ghz, machine) showing some detail - including about $20E6 in discounts on a system list price of about $37E6.

If you are a long term IBM customer and want the best possible deal, make a big deal out of testing some Sun gear for the job in the month or two before sitting down with IBM's reps to nail down your price. Spending $300K on a high end 5440/70X0 system with 100+ Sun Rays and another $100K getting everything working may seem absurd (and dispiriting to staff) - but if it makes the difference between a 20% and a 30% discount on a p6 595 + storage + DB2 etc, you'll end up saving about $2E6 net up front and have a free Sun system to0.

If you're not a long time IBM customer, I'd caution you to test your app on AIX before spending a nickel. It's a pretty good 80s Unix with 90s encrustation, but getting big apps to make effective use of big storage can be a bear.

In retrospect I should have noted that the apps test has to use realistic storage and added that using a few hundred Sun Rays to run an IBM data center is both a lot cheaper and more effective than using PCs (mainly because of licensing and responsiveness) and a good way of giving yourself some negotiating leverage.

Oddly, however, I don't care much about any of this; I've got a hard deadline on a project of different kind - and what I'm doing makes IT work seem pretty alien.

I've got 32 tons of granite mostly still on pallets in the front parking area, a September 24th deadline on getting my drystack walls built - because we're having political event here and the place will be swarming - and for the last two weeks Lethbridge, although you can normally expect about 30 sunny days in August, has had almost nothing but rain.

It's true, of course, that when you work in IT everything acquires an IT tinge - but building drystack stone walls has a lot more in common with IT that I ever thought. For example:

• It looks easy, but isn't (1)

  You're not piling bricks - every rock has six faces and they're all different. You'd think that you'd only need one nice side to face out, but in reality at last five sides have to work: the side going down has to fit what it goes on, the top side has to support what comes next, the two edges go against other rocks, and the outward facing side has to be reasonably flat.

  As a result I have rocks that look like they should fit anywhere, but which actually fit nowhere - and ugly, lopsided, mutts that sometimes fit miraculously.

  I'm told you can chip them to fit - but it's like inserting C macros in interpreted code: any idiot can do it, but it takes ten years of experience to do it well.

• It looks easy, but isn't (2)

  It's just like applications development: what you're trying to do isn't obvious and actually doing anything useful takes bitter experience - but everybody's an expert, and not only can passers by tell you're making a mess, but many feel some desperate compulsion to share that opinion.

• "expect to throw your first systems away" is good IT advice - and applies to drystack too. I build a section, decide it looks terrible, tear it down; build it again, and then my wife says it's terrible - so down it comes, again.

  Worse, what looks great up close, can look terrible from a distance.

• a great idea on paper - becomes paper mache in rainy reality.

  Since the job is to replace "railway tie" timbers used to build landscaping retaining walls while also giving the trees a little bit more root room, I thought it would make sense to pile some new top soil just above those walls before starting; so I could pull the timbers, build the new wall, and then just pull down the dirt to pack it in behind the new wall.

  We've had two weeks of rain since. Ever shovel dirt uphill with twenty pounds of mud on each boot? - and thirty pounds sticking to the shovel?

Meanwhile most of the family is enjoying sun shine and barbecue in Portland (Oregon) and only the old lady claiming the protection of some mysterious rule about matricide is here - to remind me, daily, that the thousand year old drystack walls in her native Scotland are all perfect.

But do you know what really makes it like IT? Frustrations aside, it's fun.