"Susan Hares" <[email protected]> wrote:
>
>
> Feedback to presenter.
>
>
>
> You indicated that ephemeral was dynamic protocols, but that DCHP was
> configuration protocols. The text indicates that the ephemeral state is a
> control-plane datastore,
The architecure allows for new control-plane datastores to be defined
for which the data is part of <applied>. Data from these datastores
would be tagged with an origin derived from 'dynamic'.
It also allows for control-plane datastores with data that is not part
of <applied>. Data from these datastores would be tagged with an
origin derived from 'system'.
All this is just implicit in the currect draft. We need to clarify
this.
So for the i2rs case, if you believe you need a new datastore called
"ephemeral", and you would like this data to be part of <applied>, you
can define a new identity:
identity ephemeral {
base arch:dynamic;
}
> We should be able to have 2 ephemeral datastores which are control-plane
> datastores. The tracking of the information should have enough structure to
> support it.
Yes this should be possible.
/martin
>
>
>
> Sue
>
>
>
>
>
> =======
>
> v
>
> +-----------+
>
> | <applied> |<---+--- dynamic configuration
>
> | (ct, ro) | | protocols
>
> +-----------+ +--- control-plane datastores
>
> |
>
> | +--- auto-discovery
>
> | +-----+--- control-plane protocols
>
> | | +--- control-plane datastores
>
> v v
>
> +---------------------+
>
> | <operational-state> |
>
> | (ct + cf, ro) |
>
> +---------------------+
>
>
>
> [snip]
>
> Note that the ephemeral datastore discussed in I2RS
> documents maps to a control-plane datastore in the revised datastore
> model described here.
>
>
>
_______________________________________________
netmod mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/netmod
