Re: question

[William Allen Simpson <[e-mail]>]

>From the previous two answers, I think that it is appropriate to add two
more criteria to the "subjective" list.

 1) ease of implementation / cost / time frame of adding to routers.
    - that have / don't have CLNP
    - that have / don't have ISIS
    - that have / don't have OSPF
    - that have only RIP, ARP and BOOTP.

 2) ease of transition / cost / time frame of adding to end-systems.
    - that are IP-only speakers
    - that also speak CLNP
    - that also speak IPX
    - that also speak AppleTalk

I have two hats.  One is a consultant for several of the small cheap
router projects.  These only speak IP, IPX and AppleTalk, and there is a
lot of the latter tunneling through IP.  There is no support for CLNP,
and no plans for it.  Most don't even have OSPF yet, let alone ISIS.

I have looked at the difficulty of adding SIP/IPAE, PIP, and CLNP to
this class of box, and have concluded that SIP/IPAE is the only project
feasible in the near future.  In the long term, OSPF or ISIS or
Nimrod/IDPR each requires pretty much the same implementation effort.

Under the other hat, as consultant to network users here, and as a user
myself.  The Merit/NSFnet side of my regional is a heavy touter of CLNP
and OSI in general.  Yet, the Merit/Michnet side has no support for OSI.
They are finally getting solid support for IP installed, and are busy
getting IPX and AT to IP working for the K12 constituency.  As far as I
(as a customer) can tell, there is simply no time or money to expend
completely redoing the infrastructure in the near future.  Therefore,
anything done in the next 3-5 years has to be done over IP.

Also, the regional links are 9600 or 56K, and the local links are mostly
1200 and 2400 bps.  Therefore the packet overhead needs to remain small.

Finally, the routing is exclusively static and RIP.  Any plan that
assumes OSPF or ISIS will fail in the near term.  In the long term, each
requires pretty much the same installation effort.

The only plan that meets those limitations for endusers is SIP/IPAE.


So there you have it, my request for criteria, and my assessment of the
the answers.  As it was phrased at IETF-DC, if the network collapse was
imminent, would you be willing to wait 6 months, or 2-5 years for the
implementation and deployment of each proposal?

[e-mail]