Thursday, June 24, 2010

Notes and anecdotes about moving

At the end of the day, you cannot plan enough for an IT move. You need to plan to plan, and follow those plans constantly. And you will forget things, so you write everything down- and lose the papers, or they become irrelevant the minute you write them down. You will delegate, and wonder why it wasn't done 'your way'- then, hopefully, realize that 'your way' isn't better, just your way- since you delegated it to someone competent, it's even possible they did it better than 'your way'. Nevertheless, a million funny/scary things happen.

Take, for instance, the building we moved out from. We always had power problems- a backup generator solved most of them. And the AC unit (CRAC unit in my parlance) was 27 years old- not a good place, really. So, during the inevitable wait for the pre-move backups to complete, what happens? The worst storm this area (northern Illinois) has seen in 10 years. Moving trucks were all but knocked over in our parking lot.. and, the power partially failed. The AC blew smoke waiting for the generator to come on. Saving grace was that the doors to the data center were wide open- so the fire suppression system didn't kick on. (we did that at the LAST building- it happens).

Or, in the new building, while half the team was packing the old building, myself and my lead network technician were trying to bring up the 5 circuits necessary to move in. Two DS3's.. one had bad internal wiring. 3 T1's.. of which one was dead completely from the carrier, and another, a managed circuit, meaning the carrier owns the router it plugs in to in our data center, won't come up because the carrier has decided to re-schedule the turn-up.. for a week after you move in. Plan, plan, plan. The wiring guys were onsite- the DS3 was fixed in 20 minutes, and the T1 wiring was instantly tested and assured it was in fact the carrier. Plan, plan, plan- the DS3s are redundant, and the failed T1 was part of a group of redundant circuits. The managed circuit? It just links this office to the global network, and has no redundancy- the real plan is that I sit in the group designing the replacement for that whole network- but that won't be done until next year. Too little, too late- but the right people were woken up on a weekend night in Japan- the circuit came up in 2 days, not a week.

Then there's the dumb stuff. Like, the building locks down at 6pm- you need a special card to get in after that. And our alarm isnt supposed to arm until 11pm, no problem. Except, on day 3, the alarm broke in a previously unseen way- causing it to ARM itself at random times during the day, and then go off. Careful testing formed a workaround, until the technician could get here.

Oh- and don't throw out your plans. You may need them later, filed under 'documentation'.

No comments:

Post a Comment