[IPython-dev] Multi-user IPython Notebook

[Kyle Kelley]

Thanks Brian. ;)

I'm currently running a setup where I'm subdividing a machine with .5 TB of
RAM, where each user gets 512 MB of RAM (~1000 users). Not quite ready to
show off beyond a handful of people. Email me directly if you want to check
it out.

If anyone wants specs or to talk about the environment, feel free to ask.
;) All tools are built agnostic of providers, but I am working cloud
configs that optimize for the hardware its being run on, primarily for how
/var/lib/docker is mounted.

-- Kyle



On Fri, Oct 3, 2014 at 1:57 PM, Brian Granger <ellisonbg at gmail.com> wrote:

> (Kyle, working at Rackspace, might not volunteer this information in
> this context)
>
> I highly recommend looking at Rackspace's OnMetal Servers for
> performance critical things that you want to cost contain.
>
> Cheers,
>
> Brian
>
> On Fri, Oct 3, 2014 at 11:18 AM, ssanderson <ssanderson at quantopian.com>
> wrote:
> > Kyle,
> >
> > Currently we're spinning up new containers manually over SSH using
> Fabric.
> > The container we're running is conceptually pretty similar to the
> > scipyserver container in  docker-notebook
> > <https://github.com/ipython/docker-notebook>  .  At the moment we're not
> > doing any sort of proxying, so each single user server just gets its own
> > unique URL and password, which is supplied to its user.
> >
> > As far as how many containers we run per host, we're still playing around
> > with different instance types and resource limits, so we haven't settled
> on
> > a specific number.  One of the challenges for us is that different
> research
> > activities have radically different performance profiles.  For example,
> > querying historical financial data is mostly IO-bound, analyses of large
> > datasets are often constrained by RAM, and backtesting with in-memory
> data
> > is primarily CPU-bound (in my dream world, we can farm out the CPU-bound
> > tasks to pool of worker containers using IPython Parallel, but that's a
> ways
> > off).   That said, I think our current ballpark is to run on the order
> of 10
> > containers per host.
> >
> >
> >
> > --
> > View this message in context:
> http://python.6.x6.nabble.com/Multi-user-IPython-Notebook-tp5073520p5073582.html
> > Sent from the IPython - Development mailing list archive at Nabble.com.
> > _______________________________________________
> > IPython-dev mailing list
> > IPython-dev at scipy.org
> > http://mail.scipy.org/mailman/listinfo/ipython-dev
>
>
>
> --
> Brian E. Granger
> Cal Poly State University, San Luis Obispo
> @ellisonbg on Twitter and GitHub
> bgranger at calpoly.edu and ellisonbg at gmail.com
> _______________________________________________
> IPython-dev mailing list
> IPython-dev at scipy.org
> http://mail.scipy.org/mailman/listinfo/ipython-dev
>



-- 
Kyle Kelley (@rgbkrk <https://twitter.com/rgbkrk>; http://lambdaops.com)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/ipython-dev/attachments/20141003/6122956c/attachment.html>