Last time around, the difficulty was relatively constant and the reward halved, and we saw a very smooth operation. On the other hand, while pools were a thing, I think the players in those pools were less professional, and there were basically very few data center operations.
Two scenarios are possible:
1) Difficulty is flat at the transition, which means margins are near 0% for the least efficient players. In this case, the least efficient players will be the first to turn off their rigs, because they will feel the most pain. Some of the more efficient players will keep going, because they will have high margins (people running off geothermal in iceland with cheap electricity and free cooling). Other folks will be willing to take negative margins, either because they want to throw their weight around and push weaker players out, or they are laundering dirty (criminal) money through their local power company into clean, new bitcoins. My guess is that the efficient miners + the miners willing to take negative returns temporarily + the miners doing money laundering are slightly more than 50% of the hashing power, because they are also the most profitable, so they reinvest their profits. Block confirmation times will probably spike to 20-40 minutes, then readjust after a month or two.
2) Difficulty is still climbing exponentially through the transition, which means margins are positive. In this case there will be no noticeable effects for the end user, although difficulty might start climbing more slowly.
Either way, it's not going to be a big deal. Price will probably also spike prior to the transition in anticipation of the smaller flood of new coins hitting the exchanges every day, which will push everyone more towards scenario two.