Re: [core] Conclusion -- Endpoint Client Name / Endpoint Name in RD draft

[peter van der Stok <[e-mail]>]

Hannes Tschofenig schreef op 2018-05-09 10:20:
> Hi Peter,
> 
> I have not even looked at the con parameter but I wonder how well it
> can work in practice given firewalls and NATs...
> 
Hi Hannes,

some more background.
Resource Directory was originally foreseen to handle unicast resource 
discovery within a Lowpan.
Building installations foresee that the resources and the RD are all 
located within the same administration domain
(e.g. no NATs, or Firewalls between RD and interested endpoints.)
When services administered by the RD are accessed from outside the 
administration domain, the use of DNS is recommended.
For that reason we prepare the RD-DNS-SD draft.
When the Commissioning Tool is outside the administration domain (you 
never know), access from CT to the RD also should pass via DNS IMO.
Independent of the network location of the CT, the CT can for example 
fill link-local addresses into the RD, when RD and endpoints are all on 
the same link.

I think that the best way forward is indeed an additional secure RD 
draft.
I don't know what my co-authors or Matthias think.

Cheerio,

Peter