Re: [core] 🔔 WGLC for Resource Directory

[Ted Lemon <[e-mail]>]

On May 24, 2019, at 5:24 AM, Peter van der Stok <[e-mail]> wrote:
> In managed networks which are (often) not connected to a border router, the use of a preconfigured address is recommended. 

I can’t think of a time when this would be the right recommendation.   Just because there is no border router doesn’t mean that there isn’t a service discovery mechanism.   If you have something on the network providing core RD service, then you have a “server,” and that “server” should also be able to provice service discovery.   DNS-SD is actually a pretty easy way to do this—presumably you’re already configuring a DNS server in RA or DHCP, and if not it’s easy to do, and you don’t need to write any new specifications.   The server can be very lightweight.

If manual configuration is indicated, then what should be configured is the hostname of the RD server, not the IP address, which is simply too likely to change.   This avoids the risk that you would have to go out and manually reconfigure every device on the network when you change your network configuration. 

But really, the RDA0 works just fine in this case, and that’s what I’d recommend if I were writing the document, unless you are thinking that on a network of this type, hosts are just communicating using link-local addresses.   If they are using mesh-local addresses, then whatever mechanism configures mesh-local addressing should also be able to convey the RD IP address.

> In managed networks with border routers that need stand-alone-operation, the RDA0 option recommended.

Sure.

> The use of multicast discovery in mesh networks is NOT recommended. The use of DNS facilities is described in draft-ietf-core-rd-dns-sd.

Yup.