Paul Borman <[email protected]> wrote:
> It depends on if you want to be able to just use YANG for schemas, or if
> you want to force people to use another mechanism on top of YANG to make it
> flexible enough for complete schemas.
> 
> For example, an interfaces schema may end up with paths like (I am not
> using full prefixed YANG paths, so please don't be confused by that):
> 
> /interfaces/eth0/...
> 
> But, if you want to report everything about a device, you would probably
> want:
> 
> /device/interfaces/eth0/..
> /device/protocols/tcp/...
> 
> Now step back, you actually want to contain everything about all your
> devices:
> 
> /router1.me.com/interfaces/eth0/...
> /router2.me.com/interfaces/eth1/...
> 
> And now you realize you have other things besides network devices to add to
> your schema:
> 
> /network/router1.me.com/interfaces/eth0/...
> /server/server1.me.com/memory/...
> 
> And then you want to group by continent and datacenter:
> 
> /na/dc/network/router1.me.com/interfaces/eth0/...
> /na/dc/server/server1.me.com/memory/...
> 
> And since we live in a global community:
> 
> /me.com/na/dc/network/router1.me.com/interfaces/eth0/...
> /me.com/na/dc/server/server1.me.com/memory/...
> 
> These are just examples to show the point, and the point is simple, if you
> try to authoritatively say where the root of the tree must be, either a)
> your schema cannot be incorporated, or b) you have to build something on
> top.

I agree.  This illustrates that we should not try to enforce all
models to be under /device.  If the interface model is under
/interfaces, it can then be mounted in different structural models,
serving different purposes, as shown above.


/martin

_______________________________________________
netmod mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/netmod

Reply via email to