I've used appfog and it works alright.
I am a paying customer for webbynode and like it a lot. Easy to deploy, ssh access, flexibility. $15 might be more than you want to pay, I've not tried digital ocean, but it seems like a nice entry level price point.
I like Digital Ocean (use them for http://bngal.com and http://cheergram.com) but be warned: I wasn't even able to compile nginx on the $5 plan. Kept getting memory-related errors. YMMV
http://www.cloudvps.com/ A Dutch company that I've used for the last 18 months to run my VPS without issues. It also provides VPS's to the largest online accounting software company here (Exact software) has reasonable customer support
11 Euro (14 dollars) a month for:
1GB Ram, 20 GB HDD, 250 GB traffic and a single core.
If you want even cheaper you could try hetzner.de with 8 euro/month, 512MB VPS.
Dotcloud[1] is ideal for this sort of thing (to me). If it's a production app and you're just looking for a free/cheap ride, you might need to look elsewhere, but if the complaint is that the spinup time is too slow during development, I would definitely look into Dotcloud, which lets you have development instances free, and only production apps are charged.
Might actually work for my site. I don't need a big or fast db as I will be the only one using the rails interface. External users are just served non-interactive pages.
You could check out lowendbox for cheap VPS. I had one for a year, only paid $50 for a 512MB Xen VPS. It ran a low volume rails site just fine. I did have to do a trick with mysql, and that was to disable TCP in MySQL and only use the local socket (/tmp/mysql.sock). That reduces the memory footprint of mysqld significantly.
Linode. The 20USD/month plan is guaranteed to meet your needs, and their service and support are top notch. The LISH shell is also really really useful - esp. if you are new to sysadmin stuff, and has saved my ass a lot of times. Their collection of tutorials on sysadmin stuff is another huge benefit.
When I had a rails app on Heroku (a while ago), I solved that by setting up a free Pingdom account to ping a full stack (hit the DB, hit any external services I was using) "heartbeat" route. I'm not sure if this works any more, though - they may have done something to work around it.
What's the problem with the 5-10 seconds wake up time in Heroku? This does NOT happen every time they are used, it only happens ONCE every an hour or more of inactivity. This limitation becomes problematic however if your app has cron jobs every so often, or states (e.g. node app).
It's fine for developing but I want to have this as something I can send to employers and friends. I don't want them to have to wait up to 10s to see a site that I spend a lot of time trying to speed up lol
there is a way around that. just install the new relic addon, and configure it to check your sites availability once every minute.(on a production site you want that anyway to get alerts if the app is down, and might need a restart)
like Donito said, it only needs to wake up if there hasn't been a request in over an hour -> which will never happen with the new relic addon, because it sends a request every minute.
appfog - https://www.appfog.com/
dotcloud - http://dotcloud.com
CHEAP:
webbynode, $15 entry level plan - http://webbynode.com
digital ocean, $5 entry level plan - https://www.digitalocean.com/
I've used appfog and it works alright. I am a paying customer for webbynode and like it a lot. Easy to deploy, ssh access, flexibility. $15 might be more than you want to pay, I've not tried digital ocean, but it seems like a nice entry level price point.