Monitoring just under 50 servers with them on GoGrid. Just started actually and setup has been a breeze. We ran a nagios installation for a long time but the time and effort spent on running, updating and installing it just made it not worth the effort for us.
Recommended. We monitor about 30 servers with them (Rackspace Cloud). I consider them to be one more piece of the sysadmin replaced with a SASS solution. While you certainly could setup Nagios or whatever else for monitoring all of your machines, it's incredibly easy and far less maintenance to let CloudKick handle things.
Maybe I'm too opinionated about the whole thing, but I have yet to find a cloud management solution that does things the way I want. Cloudkick, Rightscale, Chef, all seem to work ok, if you're willing to buy in to their preferred way of doing things. If you want to stray from that, though, its a lot of work, and brittle.
I don't know that they're wrong, and I'm right, exactly, but it feels like they're all trying to paper over the features and drawbacks of cloud, and making it act more like traditional hardware that people are more familiar.
We are trying to create a new way of doing things, right now we have a lot more on the monitoring side of coin; however, projects like cast are some of the next generation of tooling for CK. http://github.com/cloudkick/cast
I do not think so. We use Nagios to monitor our EC2 servers (CPU, etc.) and then AlertFox ($199/month) for external transaction monitoring.
That works well for our 25 EC2 instances (+ a few at the Planet)
We have it linked in with PagerDuty now as well.