There are a lot of reasons why I have several servers co-located at digital forest. First, they have a long history as a reliable co-lo facility. Second, I know the CTO, and trust him. Third, they charge a reasonable price relative to service. (Some people have tried to convince me to switch to Brand X, often a major national hosting company, but when I've investigated out of curiosity, I find there's no real savings, and they nickel and dime you over every last thing.) Fourth, they're local, so I can run down and work on servers in a pinch.
But there's another very important reason. They communicate. They had a small failure in their cooling system last night, and redundancy saved the day. However, they need to do some system work, and rather than just do the right thing technically--pictures here!--to ensure temperatures stay low, they also did the right thing for their customers, and informed us via email and their support blog about precisely what's happening, what might happen, and how they plan to deal with Plan B, Plan C, etc. We could use them in high governmental office right now. "Plan A has to work!" (It helps that it's 39 degrees outside their offices right now in terms of fixing the problem!)
Sunday morning, I wake up and find that one of my main Web/email servers has gone nonresponsive, probably at about 7.30 am. I check to make sure it's really down, page the on-site fellow at about 8.30 for help. He checks console and the machine is truly non-responsive. Reboots, calls me to check that I want him to proceed, and runs fschk (a Unix disk repair utility), and gets the thing back up online. Now this is typical NOC (network operation center) stuff, but I don't pay a fee for this. It's included in the co-lo charges, because they charge the right amount to allow me to get a machine rebooted every six months or so and to know that if I need help, they are there to provide it.
I have recommended them in the past, but I'll just repeat that recommendation again. d.f. has also hosted TidBITS for many years.