[Capi-bof] Charter - last call for changes

[Sam Johnston]

On Wed, Mar 25, 2009 at 2:18 PM, Michael Richardson <mcr at sandelman.ca>wrote:

> >>>>> "Sam" == Sam Johnston <samj at samj.net> writes:
>    Sam> Aside from that the "network tag" suggestion I made before
>    Sam> would allow for the creation of private networks (two
>    Sam> interfaces with the same tag would obviously be wired together)
>    Sam> but assigning meaning to those tags (subnet details, etc.)
>    Sam> would be left to the fabric. The vast majority of workloads
>    Sam> don't care, so long as they can talk to each other and their
>    Sam> clients (assuming they have any which is not always the case).
>
>  My kinds of deployments I need to do require multiple network
> interfaces per VM for security and performance reasons.
>  Typically that means 4 networks, with one of those being the Internet.
>
>  An API that does not permit me to do describe this would not be useful
> to me.  I do not believe that this is a a difficult thing to accomplish,
> and I understand it to be in scope of the proposed charter.
>

Actually this approach does allow you do to do that - you could have an
arbitrary number of interfaces and storage mounts. The question is how much
you should be able to configure those resources.

One option I was planning to put forward after the charter is nailed down is
an attribute based system, whereby we can specify basic attributes for
interoperability purposes (ip, netmask, gateway, etc.) while allowing
vendors to extend as necessary (e.g. "com.cisco.cdp: true").

If we use XML (as we arguably should) then we can follow the Atom
Syndication Format (RFC4287 <http://tools.ietf.org/html/rfc4287>) example
and use XML namespaces for extensibility and IANA for allocations (ala
Atom<http://www.iana.org/assignments/link-relations/link-relations.xhtml>Link
Relations).

The question is, do we want to go down the path of configuration and if so
how far, knowing that there is no end in sight (what about non-IP protocols?
IPv6? esoteric details like frame size?). And if we don't go far enough then
will people implement it anyway, resulting in incompatible implementations?

Sam
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.ogf.org/pipermail/capi-bof/attachments/20090325/20289c45/attachment.html