Re: CEP vs Inverse Multiplexing

Discussion in 'Cisco' started by Andre Beck, Nov 21, 2003.

  1. Andre Beck

    Andre Beck Guest

    "SysAdmin" <> writes:
    > Regarding the configuration of bundling multiple T1's to achieve a
    > higher throughput rate...
    > Is CEP okay for this?

    What does the Cisco Certificate Enrollment Protocol have to do with it?
    Either I'm off here, or you wanted to write CEF.

    > I understand that Inverse Multiplexing is the
    > ultimate way to do it (inside the CSU instead of the router)

    This would require all T1s/E1s to be octet synchronous, which isn't
    achieved as easily as it seems. At least not if the carrier doesn't
    want it.

    > but looking for an ISP to provide that I am coming up short. Everyone
    > seems to want to offer the CEP solution which I believe is per-packet
    > load-balancing.

    There is indeed CEF per-packet load balancing. It has the big advantage
    of beeing the least CPU sucking method of bundling available (except from
    indeed inverse muxing with external hardware) and is thus prefered by ISPs
    over other solutions. External hardware costs extra and is likely incom-
    patible with aggregation solutions already in use by ISPs, so they will
    not support it until they already have the hardware to do it.

    > Isn't CEP really just an aggregate bandwidth and not really "real"?

    Depends on what you call real. The big disadvantage of any equal-distance
    load sharing is that it doesn't deal correctly with all sorts of traffic.
    For instance, IP multicast will not load share on a CEF per-packet bundle.
    Another problem is, it can reorder packets. While packet reorder is a
    normal thing in the IP world and L4+ are there to deal with it, nobody
    says that the dealing with it is as efficient as not having to deal with
    it. With certain L4+, it can create significant loss of throughput as
    well as increased traffic due to packet retransmissions. So better avoid

    As long as no real physical (probably external) mux/demux is available,
    the software thing closest to that is MP. The disadvantage of MP is that
    ISPs don't like it, some just out of attitude┬╣, some because it eats more
    CPU than CEF per-packet.

    ┬╣) You know, those who also refuse to run PPP and explain you how much
    more efficient Cisco HDLC would be.
    The _S_anta _C_laus _O_peration
    or "how to turn a complete illusion into a neverending money source"

    -> Andre "ABPSoft" Beck +++ ABP-RIPE +++ Dresden, Germany, Spacetime <-
    Andre Beck, Nov 21, 2003
    1. Advertisements

Want to reply to this thread or ask your own question?

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. Ivan82
    Aug 29, 2006
  2. Replies:
  3. Multiplexing network connections

    , May 9, 2008, in forum: Computer Support
  4. Replies:
    May 24, 2008
  5. Hoffa

    Multiplexing and packet loss

    Hoffa, Apr 21, 2009, in forum: Cisco
    Apr 27, 2009

Share This Page