Using ESS on my desktop to run R in the cloud

Question

I'm interested in experimenting with writing R code on my laptop which I then execute on an Amazon S3 machine. I want the execution to be interactive because I'm building a work flow and the data is only on my EC2 instance, not on my local machine.

I could redirect X11 to put the remove ESS window on my local machine, but I've had some latency issues with that. What I would like to do is just have ESS send the execute command to the remote machine.

In Googling around I found a very interesting blog post that seems to do exactly this including redirecting graphical output to my local machine. Anyone had experience with this type of work flow on Amazon EC2 and have any tips/tricks/advice?

Answer 1

The ESS manual talks about connecting to remote R instances, not dissimilar to what the excellent tramp mode does (where you can open files via the /[email protected]:/path/file.txt generalization of /path/to/file.txt using the wonders of ssh and scp).

I have not done that. What I do a lot, though, is to start Emacs in daemon mode:

$ emacs --daemon

and to then connect via

$ emacsclient -nw    # tty, works great for `screen` over `ssh`

or if I have the bandwidth

$ emacsclient -c     # graphical via x11

The ueber-neat thing now is that Emacs keeps running, so if I start R via M-x R, that very R process also keeps running ... and I can disconnect and reconnect at will. Very powerful. You could do that on your main worker session in the cloud.

It should be possible to combine that with the remote invocation but for that we may have to ask on the ESS list for real advice.