containers development plans (July 10 version)
Serge E. Hallyn
serue at us.ibm.com
Wed Jul 11 06:41:21 PDT 2007
Quoting Andrew Morton (akpm at linux-foundation.org):
> On Tue, 10 Jul 2007 16:39:43 -0500
> "Serge E. Hallyn" <serge at hallyn.com> wrote:
> > We are trying to create a roadmap for the next year of
> > 'container' development, to be reported to the upcoming kernel
> > summit. Containers here is a bit of an ambiguous term, so we are
> > taking it to mean all of:
> > 1. namespaces
> > kernel resource namespaces to support resource isolation
> > and virtualization for virtual servers and application
> > checkpoint/restart.
> > 2. task containers framework
> > the task containers (or, as Paul Jackson suggests, resource
> > containers) framework by Paul Menage which especially
> > provides a framework for subsystems which perform resource
> > accounting and limits.
> > 3. checkpoint/restart
> I would suggest that this material be fleshed out quite a lot with usage
> scenarios, applications, etc. Something which will help the kernel team at
> large understand the *value* of this work: what it offers our users.
> Because people generally don't know that stuff yet. And if one starts
> explaining all this complexity, bug-potentiality and overhead-potentiality
> to kernel developers without first making it very clear what it will be
> gaining us, they will revolt.
> IOW: do not presume that people want *any* of this stuff at this stage.
> First up, they need to get all fired up about how cool it all will be ;)
Ah yes, great point, thanks for the suggestion.
If people who are have been working with each of these areas could send
me their use cases, that would be immensely helpful. Especially those
people who have been providing their own patches or product and know how
their customers were using them. I'm certain Cedric and probably Eric
can provide some good justification/motivation for checkpoint restart,
for example, and Kirill and Herbert for virtual servers.
More information about the Containers