The IP Over NBMA (ION) Archive

[Telford001@aol.com]

In a message dated 96-12-24 10:37:50 EST, tjsmith@VNET.IBM.COM (Timothy J.
Smith) writes:

> I was hoping that we would see a more definitive description of how
>  the ARP Client determines that the ARP Server is down (so that he
>  knows to switch over to the next in line or as discussed
>  above to potentially use an InATMARP for refresh).  I guess there are
>  some obvious ways to know (e.g. you can't establish a VCC to the server).
>  But, if you do have a VCC to the server, and the server is brain dead,
>  how do you tell?  I guess one could come up with a query for his own
>  address.  After some number of retries, he declares failure and moves
>  on.  Any thoughts?

Is this approach not bondage and discipline engineering design?
Is this issue the sort of issue that should be subject to standards?
I could probably conceive of at least 10 different ways to handle
the dead ARP server problem all equally valid.  

Are the procedures for dead router, dead mail server, dead
name server or dead dhcp server standardized by the IETF?

Joachim Carlo Santos Martillo Ajami