We are using Grid Engine 2011.11p1 (Rocks release 6.1 environment). I've
been tasked with devising a way to extend our compute cluster with cloud
services and am looking at cloudbursting AWS instances via Grid Engine.
I've seen some tantalizing evidence that others have done this but nothing
concrete. An example is:
http://www.bioteam.net/wp-content/uploads/2009/09/2009-SGE-EC2.pdf
http://wiki.gridengine.info/wiki/index.php/SGE-Hedeby-And-Amazon-EC2
but I'm having trouble finding current information on this project.
Have you built a hybrid cloud? If so, may I pick your brain? Please
contact me off-list. Here are some of the issues I'm working on
- on-demand instances vs. always-running instances
- collecting AWS billing data for each GE job
- moving user data into and out of instance
- modifying job parameters via prolog if needed
- collecting billing data via epilog
More issues will undoubtedly arise....
If there is sufficient interest and response, I will summarize what comes
up back to the list.
Thank you!
Carl
Carl G. Riches
IT Director
Department of Biostatistics
Box 357232 voice: 206-616-2725
University of Washington fax: 206-543-3286
Seattle, WA 98195-7232 internet: c...@u.washington.edu
_______________________________________________
users mailing list
users@gridengine.org
https://gridengine.org/mailman/listinfo/users
