[TriPython] Pangeo demo

[Calloway, Chris]

I found out why the Pangeo demo did not go so well in my lightning talk last night:

“We've been struggling to get autoscaling going for the cluster and
have not yet succeeded.  So currently the cluster is scaled way down
to avoid excess charges unless we know we need it scaled up.   So if
you want to do a demo, let us know when, and we can scale it up for
your demo, okay?”

But you can watch Rich Signell of USGS give an updated demo of the same notebook:

https://www.youtube.com/watch?v=UQqWHeUQHEQ

The mind-blowing parts are at 10:00 and 13:10.

There is also an interesting bit at the end about dask_jobqueue allows you to distribute Dask on a cluster job with queuing systems like PBS, Slurm, MOAB, and SGE, which used to be really hard to work around.

Even without your own cluster, you can see the wonder of Datashader interactively by running this notebook on Binder yourself, no Dask needed for this demo:

https://mybinder.org/v2/gh/reproducible-notebooks/Tides_TriMesh_Datashader/master

This is a Datashader display of a tidal dataset. If you zoom in around Puerto Rico, you can see the data being resampled. There are 3.7M triangles in the tidal dataset.

--
Sincerely,

Chris Calloway
Applications Analyst
University of North Carolina
Renaissance Computing Institute
(919) 599-3530

-------------- next part --------------
   I found out why the Pangeo demo did not go so well in my lightning talk
   last night:



   "We've been struggling to get autoscaling going for the cluster and

   have not yet succeeded.  So currently the cluster is scaled way down

   to avoid excess charges unless we know we need it scaled up.   So if

   you want to do a demo, let us know when, and we can scale it up for

   your demo, okay?"



   But you can watch Rich Signell of USGS give an updated demo of the same
   notebook:



   [1]https://www.youtube.com/watch?v=UQqWHeUQHEQ



   The mind-blowing parts are at 10:00 and 13:10.



   There is also an interesting bit at the end about dask_jobqueue allows you
   to distribute Dask on a cluster job with queuing systems like PBS, Slurm,
   MOAB, and SGE, which used to be really hard to work around.



   Even without your own cluster, you can see the wonder of Datashader
   interactively by running this notebook on Binder yourself, no Dask needed
   for this demo:



   [2]https://mybinder.org/v2/gh/reproducible-notebooks/Tides_TriMesh_Datashader/master



   This is a Datashader display of a tidal dataset. If you zoom in around
   Puerto Rico, you can see the data being resampled. There are 3.7M
   triangles in the tidal dataset.



   --

   Sincerely,



   Chris Calloway

   Applications Analyst

   University of North Carolina

   Renaissance Computing Institute

   (919) 599-3530



References

   Visible links
   1. https://www.youtube.com/watch?v=UQqWHeUQHEQ
   2. https://mybinder.org/v2/gh/reproducible-notebooks/Tides_TriMesh_Datashader/master