Regardless of which pool you are in, you should always set up backup miners. I love that BTC Guild has multiple servers because then my backups can be on BTC Guild itself! In my case, I had each card hashing on UK and NL, so when UK went down, they automatically switched full-speed to NL. Also, if one server has idle issues, the other worker picks up the slack.

This is quite easy to do. You simply start two workers per device. You can prioritize them if you wish with poclbm/-f or Phoenix/AGGRESSION. I have mine running on Phoenix with equal aggressions, so it splits 50/50. On the other hand, the frames trick for poclbm will allow you to give one pool 99%+ of your hashes and another pool < 1%. If Pool A goes down, Pool B gets 100%.

Even the biggest pool has had connection errors, DDoS, and other problems, so the responsibility is on you to prepare things wisely on your end so that everything is smooth and automated. And the pool operators will do everything they can as well, of course.

For me, especially since I didn't have any real downtime, our recent luck more than makes up for any complaints:

Luck in past 24 hours       724013 shares   (+21.2%)