We can look at all accounts with one dashboard and save a ton of money by having awareness of how we're spendingSome of the UI is a little clunky. They do listen when I tell them about this, but one thing that drives me insane is that it defaults to a cumulative view of how much you've spent, as opposed to day-to-day. Even when you've selected day-to-day, you have to turn off a cumulative switch. They've done some work on fixing that in the UI, but I don't know why they're thinking that way and why they think people would want to look at it that way. The one thing I wish it had was that it worked more closely and more cleanly with custom data inputs, which I know is a lot to ask. It could probably have a nice warning page for all the things. Not for me, not for the big administrator, but for all my users to whom I give accounts so that they can look at their own things. It would be helpful if they had a landing page that just said, "Hey, you're losing money here. You're possibly losing here or there." It expects you to have a certain amount of knowledge of what you're doing when you first come in. It does have a dashboard, but it's very esoteric. Something just right on top that says "You're bleeding cash right here," and a graphic would be nice. If I want to set up a new account, there's no API. It's all done by hand. It would help us in the deployment of new users if there was a new user API that we could call with a program, as opposed to doing it by hand. We would like to be able to report how busy our machines are so that if a machine has been sitting there idle all week, we should turn it off. Sometimes, it's not just how busy the processor was, it's also about how much RAM was being taken up when that thing was running. Often, it can run low CPU and high RAM, and we still have to keep it running because of that. There is this arcane workaround to get that to work by installing something special on the individual machines one by one. Deploying that way is very arcane. I wish there was a better way of doing that. It seems it works with CloudWatch, at least on AWS, or CloudTrail, to do it, but I don't know for sure. They probably would've done it if it was easy, but it would help a lot if, in the deployment, it also was aware of memory usage, not just CPU usage by default.
