I was amazed at how many people showed up for the inauguration. Here in DC the roads were closed, the weather was cold and the atmosphere was in a word, electric. I also witnessed some disturbing things, like people who had arrived from hundress of miles away but could not get in, and many more who did get in but had to watch it all on the jumbo-tron screens anyhow. Still, being there is being there, and nobody can take it away.

And this is a segue into my observations on some interesting patterns of humanity that align perfectly with our solution domains, so follow me on this. As I sat in a hotel room on Tuesday, debating whether I should brave the extreme cold and traffic, I am reminded that one of our presidents (Harrison) gave a too-long inauguration speech in such conditions, developed a cold and died of pneumonia a month later.  As for me, I had a problem to solve - when I turned on the morning news I found a common "of scale" ingress and egress scenario, something I predicted but was stunned as to the magnitude of its reality. Yes, an important day for America and for the world - and worth at least a superficial examination of the logistics of people-movement.

I noted that the cabs, trains, buses public and private, and other transportation mechanisms had been trickling people onto the scene for days, increasing by the hour. The DC train systems had dropped off over 200k people in a matter of hours, probably setting records as well. But when we think of the drop-off mechanics, these are not people arriving in-bulk. They are collecting from various places in a near-transactional manner. A handful here, a handful there, each one patted-down for their belongings one-at-a-time no differently than our transactional stored procedures check our inbound data one-at-a-time. Then the festivities started, the regalia of our peaceful transition of power, and when it was all done, we saw another interesting effect. It was time for everyone to make an orderly exit.

So people who had arrived early in the day and had likewise confirmed a front-row position for the festivities -  now did an about-face to exit - only to find a sea of humanity between themselves and the exits. I use the word "exits" here loosely, because they would not be patted-down to leave like they had been to enter, so the egress was a bit smoother and more steady. Uh, you know, like we pat-down the data when it first enters our warehouse, and deliver it clean on-demand. Ohhh, the parallels.

Discussing this at the watercooler, a couple of colleagues wondered out loud how to make a mass-egress work. How would we (safely) empty the Mall of the majority of people in a short period? After all, some of the attendees didn't make it back to their hotels until almost nine hours later! One person suggested to do it "the Netezza way" by providing 100 helicopter pads, plus 400 helicopters, each of them alighting on the pads every five minutes with a 20-minute round trip to spirit people out. This would leverage the vertical space, not just the horizontal. But this model won't serve, since the helicopters waste the return leg. One of them suggested conveyor belts, objecting to the Netezza way. But I suggested that this better represented the Netezza way, a streaming model of constantly moving data. The helicopters could move people at only 1/100th the speed of the same number of conveyors, and they wouldn't have to move all that fast.

The streaming model is something that shakes the rafters on our reporting models, but as with any problem of scale, we must provide the physical plant first, and it has to address the problem on purpose.