[c-nsp] Multilink PPP over Frame and ATM

Charles Sporkman spork.sporkman at gmail.com
Fri Feb 10 22:32:04 EST 2006


Hi all,

This one's kind of stumping me...  We're a Covad partner, and
sometimes it's most economical for everyone involved to go with a
Covad T1 for some customers.  Covad buys a T1 loop from Verizon and
puts a standard T1 router on the CPE side and they talk frame relay
out the T1.  The rest of Covad's DSL network is ATM, so they do a
Frame <-> ATM conversion.  Our backhauls to Covad are ATM T3s.

So in this era of ever-increasing bandwidth use, we've got people
wanting more than 1.5 Mb/s of symmetric bandwidth.  For straight T1s,
there's plenty of options.  For these Covad T1s, I wasn't seeing a
good way to truly "bond" the T1s.   Then I saw that Speakeasy was
doing a big promotion on Covad T1s and they were speaking of
"bonding".  I found this document on their site:
http://www.speakeasy.net/pdf/Speakeasy_BondedT1_WhitePaper.pdf
On the bottom of page 6, they talk about Mulitlink PPP over Frame and ATM.

So that looked like the answer.  I've got two Covad T1s on order into
the office.  One is done, the next should be ready Monday.  I've got a
2620 with two T1 wics in it.  The other end is the trusty 7206.

I used the two following docs to setup both ends.

frame:
http://www.cisco.com/univercd/cc/td/doc/product/software/ios124/124cg/hqos_c
/part30/ch05/qslfifr.htm#

atm:
http://www.cisco.com/univercd/cc/td/doc/product/software/ios122/122newft/122
t/122t13/ftatmmlt.htm

Here's where I'm confused.  I have this up and running with the single
T1 and it seems to work.  But in both of those docs they talk about
this multilink setup being strictly for QoS.  This overview doc
(http://www.cisco.com/warp/public/788/voice-qos/designing-mlp-over-fr-atm.html)
even states that you cannot use this to bond two circuits together. 
So how is it "multilink" then?  I understand that I'm layering PPP
over everything to get control of fragmentation and QoS away from the
underlying transport, but it seems like a misnomer.

And of course Speakeasy claims to be doing it with Cisco 18xx CPE and
Juniper aggregation routers and that confuses things even further.

Am I wasting my time trying to do this?  The docs are real fuzzy on
the part I'm interested in.  QoS is not even a concern for these
customers.

Anyone doing anything remotely like this?  Any ideas?  I'll know if it
works sometime next week, but I'm really wanting to understand the
concepts here...

Thanks,

Charles

Here's a snippet of both ends of the config.

CPE:
!
interface Serial0/0
 description covad circ 103-341-xxx vz circ 32HCFU.239xxx hse 396
 no ip address
 encapsulation frame-relay IETF
 no fair-queue
 frame-relay lmi-type ansi
!
interface Serial0/0.1 point-to-point
 description frame pvc
 frame-relay interface-dlci 16 ppp Virtual-Template1
!
interface Serial0/1
 description covad circ 103-336-xxx vz circ 32HCFU.227xxx hse 386
 no ip address
 encapsulation frame-relay IETF
 no fair-queue
 frame-relay lmi-type ansi
!
interface Serial0/1.1 point-to-point
 description frame pvc
 frame-relay interface-dlci 16 ppp Virtual-Template1
!
interface Virtual-Template1
 description ppp template for MPoFR
 bandwidth 3072
 ip address 10.1.1.2 255.255.255.252
 ppp multilink
 ppp multilink interleave
!

7206:
!
interface ATM4/0.2614 point-to-point
 pvc 0/2614
  encapsulation aal5snap
  protocol ppp Virtual-Template1
 !
!
interface ATM4/0.2615 point-to-point
 ip unnumbered Loopback8
 pvc 0/2615
  encapsulation aal5snap
 !
!
interface Virtual-Template1
 description test MPoATM office circ
 bandwidth 3072
 ip address 10.1.1.1 255.255.255.252
 ppp multilink
 ppp multilink interleave
!
mppp-test-568#ping 10.1.1.1

Type escape sequence to abort.
Sending 5, 100-byte ICMP Echos to 10.1.1.1, timeout is 2 seconds:
!!!!!
Success rate is 100 percent (5/5), round-trip min/avg/max = 12/15/16 ms
mppp-test-568#



More information about the cisco-nsp mailing list