On Wed, 20 Jan 2010, Jesse Keating wrote: > I'm a bit more interested in setting up my own clouds, rather than using > existing ones. But we would still have to generate stuff and work with > a cloud in any case. So this is a complicated question -- one that I think deserves some frank discussion. We have a project called oVirt, but no one is actively working on it, because... We've got a product called RHEV-M that came from the Qumranet acquisition, but it's not yet open source. A large team is actively working on this problem, but it's going to take them a while. Maybe quite a while. We've got a project called Deltacloud, the purpose of which is to create an abstraction layer around cloud management, much as libvirt creates an abstraction layer around virt management -- and drivers still need to be written for both public cloud providers and private cloud tools. One of the reasons for understanding EC2 really well is to gain a deep understanding of the use cases for Deltacloud; a useful goal is to manage all Fedora/community EC2 interactions with the Deltacloud API instead of the Amazon EC2 API, and then test our ability to move workloads from EC2 to, say, Rackspace or IBM or whomever. We've got some homegrown stuff in Fedora that might be useful for management of private clouds -- I know mmcgrath has some tricks up his sleeve. Some subset of that functionality could sit beneath Deltacloud drivers -- it might be worthwhile for someone to start looking at that. None of this stuff is yet critical path, IMO -- but once we've got "Fedora for EC2" solved, those questions become immediately Very Large. And right now, we've got way more questions than answers. I hope this list is a place where we can talk through some of this stuff. --g -- Computer Science professors should be teaching open source. Help make it happen. Visit http://teachingopensource.org.
