Still looking, searching, reading for some way to have good, decent and often backups of the machine because if the power goes down or something happens, when the instance goes down .. you lose the data inside the instance.

Saw that there are threads about this on amazon's aws ec2 forum. Amazon's staff say that they are thinking about this and how they could provide either a database service (mysql or something else) or some kind of really useful and permanent storage (like the 160GB /mnt partition, but that doesn't wipe when your instance goes down).

The service that is already available and everybody is expecting to provide persistent storage is Amazon's S3 ( Simple Storage Service ). It's kind of a huge and cheap storage service, a walkthrough is available here. The problem for the time being is that there are no stable linux implementations.

A solution would be the s3 / fuse interface. But that seems to be in alpha state now (btw, huge thread on aws about this). This could be a solution to many problems ... but it will take a while until it's done.

Another project in alpha state is s3DAV. Maybe using this and davfs2 could be another way.

I'm not a backup guru so i'll just keep reading, trying to find out as much as i can but in the mean time i'm thinking of making a system snapshot as an image every 24 hours ,uploading it to S3 and keeping the most recent 4 snapshots as this would be appropiate for my new system (the compressed image has around 400MB and of course besides the blog the server is empty; this wouldn't be ok on a 30GB database server or whatever live system with some amount of data, but for me it's ok now). I have attached a script i made for this purpose.

This is only best practice for now and has too many downsides for it, but until i figure out something better i'll stick to this.

Another simple way would be to make some scripts that once an hour do a mysqldump --all-databases , copy the /var/log/* and upload this on S3 and have some scripts that when the instance boots, download the latest backups from s3 (the database part .. because the logs to be copied back are pointless) .
I'm trying to achieve this without using other machines or other services from other providers, because if i did this, the whole picture would change, and i wouldn't be using just EC2 and S3, but other third party services and cost would go up and i would start thinking about not using EC2.