Automated Openstack Installations with Chef, Developers, and Sysadmins
Abstract
For educational organizations with existing, well supported IT infrastructure and skilled staff, running Openstack for local cloud services presents a number of advantageous over using commercial providers. However, running... [ view full abstract ]
For educational organizations with existing, well supported IT infrastructure and skilled staff, running Openstack for local cloud services presents a number of advantageous over using commercial providers.
However, running Openstack effectively requires an agile operations team committed to infrastructure automation. Since 2011 Information Technology Services at the University of Toronto Libraries has been using Chef, an open source configuration management application, to
control its IT infrastructure. When we received an Ontario MTCU grant in 2014 to implement a large scale academic cloud storage and computing platform (Canadian Text Archive Centre: CTAC), using Chef with Openstack was the obvious choice.
This presentation discusses the challenges and benefits associated with using Chef to apply software development methodologies to Openstack installation and management.
Challenges include:
-using the largely undocumented and occasionally buggy Chef community
cookbooks for Openstack
-significant investment of staff time before results are widely visible
-performing Openstack version upgrades in a centrally controlled environment
Benefits include:
-fully and easily repeatable Openstack development and production
environments
-the certainty that servers will not stray from defined configurations
-established and sophisticated tools, methodologies, expertise associated with the cutting edge of open source software development
-distribution of work between developers and sysadmins via the Devops mindset, allowing greater diversity of project input
Takeaways include:
-there are powerful benefits to using Chef to install and manage Openstack
-skilled sysadmins and developers are a prerequisite
-initial work is significant (3 months to build development environment)
-subsequent work is minimized and automated (3 days to build production environment)
Authors
-
Graham Stewart
(University of Toronto)
-
Chris Crebolder
(University of Toronto)
Topic Areas
Advanced Research Computing (ARC): Canadian leadership in international research projects , Advanced Research Computing (ARC): Innovations in platform / portal tools & software devel , Advanced Research Computing (ARC): Innovations in computational research (i.e. software, s
Session
HPC2.2.1 » Advances in Cloud Computing (10:00 - Tuesday, 21st June, CCIS 1-140)
Presentation Files
The presenter has not uploaded any presentation files.