Ipython notebook provides a interactive and intuitive way to manipulate with data. As I am thinking to playing with larger size data, my hands are tied due to the limitation of my laptop. Then I realize I’‘ve had the access to the Purdue server Hathi, which is mainly for Hadoop job. More excitingly, it contains all packages from Python (including anaconda distribution), MPICH2, and R. (I found this when I am trying to install the Rmpi package.) The inspiration to use ipython hosted by Hathi hits me immediately.

The steps are straightforward,

  1. Make sure your can login without password. This can be done by following reference 2.

  2. Start ipython on Hathi.

% on server
module add anaconda
ipython notebook --no-browser --port=7777
  1. Setup the ssh tunnel from the local machine to the remote machine.
% on local machine
ssh -L 7777:localhost:7777 user@hathi.rcac.purdue.edu
  1. Access remotely hosted ipython notebook through your favorite brower at localhost:7777

As Joe Hamman said,

the beauty of this is that I interact only with the browser on my local machine, while having access to the computing power and file system of the remote machine.

Reference

  1. http://www.hydro.washington.edu/~jhamman/hydro-logic/blog/2013/10/04/pybook-remote/
  2. http://www.linuxproblem.org/art_9.html