[R-sig-hpc] Distributed computing

[Whit Armstrong]

It is incredibly easy to get started w/ ec2.

1) pick a base ubuntu from the link that Dirk sent, or a redhat based
distro if that's your thing
2) boot up a micro instance (2 cents per hour, or free for new accounts).
3) ssh into the box
4) install all the custom R packages you want and any additional
software w/ apt-get or yum.
5) when you're happy with the installed packages and setup, halt the
box (but do not terminate it)
6) in the management console, right click on your instance, and choose
'Create Image (EBS AMI)'
7) once this process is completed you have your own custom spin with
its own ami identifier

For the deathstar image I built on Friday, this took about 20 minutes.
 Far less if you're only installing from apt-get or yum.

Additionally, you can try the AWS.tools package I posted to cran last
week, which allows launching and terminating instances, and s3
commands from R.

http://cran.r-project.org/web/packages/AWS.tools/index.html
or
https://github.com/armstrtw/AWS.tools

fyi, my public image for deathstar is under this ami: ami-9d5f93f4.

Cheers,
Whit


On Sun, Oct 16, 2011 at 10:48 AM, Dirk Eddelbuettel <edd at debian.org> wrote:
>
> On 16 October 2011 at 09:19, Sean Davis wrote:
> | On Sun, Oct 16, 2011 at 8:56 AM, Zachary Mayer <zach.mayer at gmail.com> wrote:
> | > Could you provide a link to some of these AMIs?  The only only one I'm
> | > aware of is the bioconductor AMI.
> |
> | The bioconductor AMI would be a reasonable choice I think.  Here is
> | another that bundles R:
> |
> | http://cloudbiolinux.org/
>
> Whit (who is leading a parallel thread here under the 'deathstar' banner
> covering his R-based AMI tools) and I discussed that in email a few weeks
> back.  No clear choices dominate but ...
>
> | Dirk may be aware of others.  Dirk (I think) also maintains a ton of
> | R-related debs, so you could probably get away with a pretty standard
> | debian AMI and simply add R via the normal apt get to get what you
> | want.
>
> ... Ubuntu has builds at http://cloud.ubuntu.com/ami and via another link on
> the EC2 tutorial page at the Ubuntu wiki, and Eric Hammond has a popular set
> too with a server configuration at his http://alestic.net site.
>
> I guess curerntly good approaches are to either start from the BioC one and
> strip it down (if you're like and don't do Genomics) or start with the basic
> Ubuntu ones and add packages as Sean suggests. That is the route that Whit
> took.
>
> Dirk
>
> --
> "Outside of a dog, a book is a man's best friend. Inside of a dog, it is too
> dark to read." -- Groucho Marx
>
> _______________________________________________
> R-sig-hpc mailing list
> R-sig-hpc at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-hpc
>