Hello all,
I've been trying to find a way to bond multiple T1s together, and had basically written it off as not doable in our specific situation. In short, we're selling Covad T1s, so I'll give a brief overview of how they deliver that service:
-CPE is set for Frame Relay encapsulation
-Covad buys T1 loop from LEC, terminates it into Nokia DSLAM
-DSLAM converts FR to ATM
-DSLAM does NOT have the ability to do bonding at the ATM level
-The PVCs for each T1 come to us over one of our ATM T3s to Covad
We did not want to incur the load of doing load-sharing w/CEF, and PPPoE seemed like an inefficient (and clunky) way to try and bond two or more T1s.
I recently saw Speakeasy offering a bonded T1 solution using Covad as the last mile provider. Looking at their white paper on the subject, I found they were doing something kind of interesting.
On the CPE (Cisco 18xx), they run "MLPPPoFR" or "multilink ppp over frame relay", and then on their Juniper device (ERX) that takes the ATM handoff from Covad, they run "MLPPPoA" or "multilink ppp over ATM". This sounds interesting.
However searching CCO for either "MLPPPoA" or "MLPPPoFR" is only giving me three hits that only tangentially mention those protocols. Google (web/groups) offers a single page as well with no good links.
What am I missing here? Does Cisco call these two encapsulations something different?
Is anyone here familiar with this? What kind of cpu overhead would the MLPPPoA protocol generate?
Thanks,
Charles