Re: [manet-dlep-rg] DLEP session establishment

["Taylor, Rick" <[e-mail]>]

> From: [e-mail] [mailto:manet-dlep-rg-
> [e-mail]] On Behalf Of Teco Boot
> Subject: Re: [manet-dlep-rg] DLEP session establishment
>
> This could work.

Excellent, I see consensus approaching... ;)

>
> It removes the additional multicast message, and state.
> The Peer_Discovery is some kind of ignition packet. Could be send with
> link local multicast every now and then (say, every 5 seconds), to
> discover recently connected devices (Ethernet port state is not reliable
> here). A remote server could send the Peer_Discovery packet with its
> global unicast address. Then, the modem will set up the TCP connection.

Yes, just what I thought.

>
> Sending the multicast from router to modem (and having the TCP server on
> router) adds some complexity on the modem, in that this multicast packet
> shall not be forwarded over the modem link (e.g. RF path). Cannot be done
> with L2 MAC filter, as this would block a set of multicast addresses. The
> filter has to block the assigned IANA DLEP multicast address.
> LLDP better fits our requirement for discovery. It doesn't take away the
> need for the multicast Peer_Discovery. On the other hand, LLDP is not
> widely implemented, I think. And would be bridged on modems that doesn't
> support it.

I agree, but I imagined Peer_Discovery being link-local multicast/broadcast.  You are right that a multi-hop scoped multicast is a nightmare.

And yes, perhaps we should be using mDNS/Bonjour for discovery rather than re-inventing the wheel here.

Should we put in some text: "Unless there is an alternative discovery protocol in use, such as a-priori static configuration or mDNS, then Peer_Discovery messages SHOULD be sent every X seconds to the link-local multicast address"

>
>
> On 2: Ident TLV could be included here (e.g. for logging). Port would be
> removed. Maybe add mandatory authentication TLV (with as default a blank
> password, this helps getting it implemented). TCP address is in header.

I disagree.  Ident is only needed when the TCP connection is made, and it keeps the UDP packet size down by not including it.  Implementations shouldn't log Discovery packets, they will be constantly bouncing around.

>
> On 4: ALL supported TLVs are send, not only the mandatory.

Yes, I stand corrected.

>
> On 5: Add all supported TLVs? Remove Ident when sent on 2?

Why does the router advertise it's supported TLVs?  I thought we had agreed that the router should ignore any TLVs it doesn't understand, and we haven't yet discussed the router 'demanding' TLV support from the modem.

>
> Add a 6, with Peer_Initialize_Ack from modem to router? This gives the
> modem a change to terminate the session before it was active.

Not sure it is needed, unless the Router can change the 'parameters' of the session in its ACK, the modem already knows everything it wants to know, so we can avoid an extra step.

Rick
The information contained within this e-mail and any files attached to this e-mail is private and in addition may include commercially sensitive information. The contents of this e-mail are for the intended recipient only and therefore if you wish to disclose the information contained within this e-mail or attached files, please contact the sender prior to any such disclosure. If you are not the intended recipient, any disclosure, copying or distribution is prohibited. Please also contact the sender and inform them of the error and delete the e-mail, including any attached files from your system. Cassidian Limited, Registered Office : Quadrant House, Celtic Springs, Coedkernew, Newport, NP10 8FZ Company No: 04191036 http://www.cassidian.com