1. How do you "increase the amount of RAM" on an EC2 instance other than spinning up a new larger instance?
2. Why didn't they use EBS and take frequent snapshots to S3?
It sounds like they were running an instance store EC2, then fired up a larger instance and stopped the first one? This has to be a textbook case of how not to use EC2.
You can't. Instance types are fixed. He was likely using an EBS-root configured AMI, where the mount point for EBS is /. These are EBS-backed, but since the EBS block is automatically created when you spool up the instance; it is also automatically deleted when you terminate the instance, unless you specify otherwise.
This is not actually true. If you Stop an EBS-root EC2 instance, the EBS Volume can be safely detached and moved along to other instances, or you can restart the instance and your data will be ok.
Additionally, There's no excuse for not snapshotting your EBS volumes of any sort, and these snapshots persist beyond the lifetime of the volume.
I find it just a little sad that relatively few (less than all) EBS customers don't snapshot their volumes.
You do not have to move your volumes: while the instance is in a "stopped" state you can change it's instance type and then restart it. EBS-root let's you change hardware as fast as you can reboot with just three commands.
You are, of course, correct. Notice in my post that you replied to I was careful to use the word terminate, not stop, when referring to what I assume he did to his instance.
edit: After re-reading my post it probably wasn't clear to people who don't use EC2; so your reply is probably for the best. :)
I have to repeatedly tell people that EBS-backed is not persistent. Amazon did everyone a big disservice by calling something -backed when it was actually transient.
persistent until you terminate the instance.
Unlike a EBS volume that you explicitly attach, which persists beyond the termination of the instance you attach it to.
2. Why didn't they use EBS and take frequent snapshots to S3?
It sounds like they were running an instance store EC2, then fired up a larger instance and stopped the first one? This has to be a textbook case of how not to use EC2.