IP over SONET considered harmful?

Mailman List Mirror (Read Only)
1

Perhaps.

  I am concerned about the growing movement towards IP over SONET.

  Previously in my career I was a vocal advocate of IP over ATM for
  several reasons, primarily traffic engineering and statistical
  gathering ability (obvdisclaimer, this required an autonomous
  unshared network used only by the ip provider for interhub
  traffic).

  However, I am firmly rooted in the bandwagon advocating IP OVER
  SONET FOR EVERYONE. Firmly.

  Accordingly, I am concerned about the visible L3 hop inherent to
  packets transiting routers.

  An ATM core is, of course, invisible to L3; so the number of
  switches or hubs through which a packet travels is inconsequential
  to the TTL of the packet.

  When a backbone is constructed with a PACKET over SONET core, the
  packet is likely to decrement the TTL by 2 at every hop. The
  number 2 is assumed because you are likely to leave from a router
  different than the one you come in.

  Since I tend to think in formulas, I'll encourage you to do so as
  well.

    Variable Meaning
  -------------- ---------------
  ROUTER L3 device which decrements
        the ttl of an IP packet

  TRANSIT_HUBS The number of hubs which neither sources
        nor delivers the packet

  NONCOREROUTERS The number of routers which accept
        or deliver traffic to a peer or customer

  TRANSIT_ROUTER A router which transits the packet

  TTL_DECREMENTS The number of ttl counters which
        this network decrements

  Assuming an architecture with dual core routers and two layers of
  hierarchy (backbone v. customer aggregation/peering), I believe
  the following formulae dictate the TTL degredation expected:

  ATM NETWORK:

2

Yes.

For the very reason you have identified of TTL decrement on hop to hop links
on IP over SONET we are in the process of building an optical Internet where
we will use WDM cut through (and maybe SONET label switching) to provide the
best performace possible.

With WDM cut through there are absolutely NO switch or router latencies.

For more information please see the CANARIE web site at

http://www.canarie.ca

Bill

Bill St. Arnaud
Director Network Projects
CANARIE
bill.st.arnaud@canarie.ca
http://www.canarie.ca/bstarn
+1 613 660-3497

3

  I consider Windows 95 to be the least common denominator,
  which has a default IP TTL of 32. Yes, 32. So that implies
  that each NSP should decrement no less than 8 TTLs.

That's broken. I quote from RFC 1340, dated July 1992:

   The current recommended default time to live (TTL) for the Internet
   Protocol (IP) [45,105] is 64.

This does not change reality, of course, but it also does not
make it less broken.

- H�vard

4

> I consider Windows 95 to be the least common denominator,
> which has a default IP TTL of 32. Yes, 32. So that implies
> that each NSP should decrement no less than 8 TTLs.

That's broken.

  Well, perhaps, but my wording certainly is; of course I mean 'no
  more than 8 TTLs'.

I quote from RFC 1340, dated July 1992:

   The current recommended default time to live (TTL) for the Internet
   Protocol (IP) [45,105] is 64.

This does not change reality, of course, but it also does not
make it less broken.

  Arguable; not following a recommendation is not broken. It's just
  dumb.

  I'd like to see IETF make 64 a requirement or standard. But of
  course that is painful.

  -a