Usually yes, but that is why you take advantage of specialized CPU instructions for bulk loading and operating on data. From what it says that is part of the optimization that these folks are taking advantage of (see comment mentioning SSE instructions).
That is a correct understanding, what Kloudless does is provide a uniform interface for 9 different services. So rather than writing the same code slightly differently and/or having to learn 9 different SDKs and dealing with these different interfaces and managing the different tokens (some sources use Oauth1.0a, others OAuth2) you simply tell kloudless which account to access and after your users authenticate you can access it in the same way no matter which storage service they choose.
If you just have some application data that you are looking to back up, it is pretty straight forward to back stuff up to Amazon's S3 or glacier (depending on your usage of said backups). Storage is relatively cheap and the amazon api is pretty when using either s3cmd or boto.
The nice thing about Vagrant is that it is more platform independent, so even if you are doing your dev work on a mac you can still use the same dev-box as anyone else. Docker, on the other hand, is linux only (and requires a certain kernel version) to work well. Not to say you can't use Docker, and they seem to work well together, but they are kind of solving different problems.
I have checked out saltstack (not in too much detail, since I have been busy with puppet). But I wasn't sure if there was an equivalent to exported resources in saltstack (I want to use them for automated monitoring and orchestrating of other services), since that is a pretty useful thing and I was somewhat skeptical about having everything defined in YAML, since there are somethings that I think one wouldn't be able to do... but I have been meaning to look into it!
Oh yeah, that is true, but the point is to learn how to do it by hand so you can do more than just bring up a php app (like develop django or rails or anything else).
Yeah, puppet isn't the easiest thing to get started with, but the primary thing I wanted to use puppet for was to make developing my puppet manifests faster/easier. It has helped a bunch, since before I had to deploy to a fresh EC2 instance every time I wanted to test out a fresh build.