2 T-1's, act as one big T-1, 1 router

[brian2112]

I have one router, 2 ethernet ports, 2 serial, 2 T-1's. They are asking me if I can set these 2 T-1's to act as one big T-1. what do I put in the router script to do this??

 

[Belushi]

What type of router do you use? If you use Cisco routers there is two ways to do but I beleive your ISP will have to work with you to support both solutions. Solution number 1 is using Cisco Expres Forwarding (CEF). This is a proprietary Cisco protocol and thus both your router and the ISP's gateway router must support it. Option 2 is multi-link PPP which I believe is an industry standard protocol. Both are relatively easy to configure. If you were to set this up you would see the exact same level of traffic going over both T1s at any given time. Thus giving you the impression of a 3 MB pipe.

 

[jeter]

ppp mutilink is what we have used in the past to bined two circuits together. Works well. “Reserve your right to think, for even to think wrongly is better than not to think at all”

Fisher CCNA

 

[brian2112]

It is a Cisco 3600 Series Router.

 

[multilinkguru]

yes, multilink is a good way of doing it, if not using the ip load balancing.

But I am not sure if cisco mlppp is robust, we have got it to crash in our lad many times. It doesn't even fragment a frame. It just sends out a frame at a time on each T1.

I am not selling any box, but I work for Tasman Networks and we pioneer in multilink routers. Both PPP and frame relay. Yes, they are all standards based and inter-operable.
Fragmenting a large frame into 2 halves, sending each fragment to each T1 simultaneously actually gives true multilink and big-fat-pipe performance, as the whole frame now takes half the time to reach the other end.

good luck!

 

[tcvander]

Who is your frame provider? Some providers have inverse mux's that they will supply you for this. Todd VanDerwerken, CCNA, CCDA
Technical Consultant

http://www.ciscolinks.net

"If at first you don't succeed...then sky diving isn't for you!"

 

[multilinkguru]

from what I know, inverse muxs are proprietary implementations and need the same manufacturer's equipment at both ends. Multilink standards (MLPPP and MLFR) were evolved to make multiple vendors work together.

I suggest you do either MLPPP or MLFR, depending on what your provider supports. If your provider doesn't support any of the ML protocols, then ip load balancing is the option, you have to burn a lot of ip subnets, complicated configurations, and still, fat-pipe bandwidth is not guaranteed.

 

[tcvander]

Your statement is true, but that is why I stated that stated that some providers have the mux's that they will provide you at both ends.
If the goal is to have a 3M pipe, then the only option is inverse mux or MLPP or MLFR (if the provider supports it).

Keep in mind the following:
3620 R4700 80MHz RISC
Fast Switching - 20-40kpps
Process Switching - 2kpps

3640 R4700 100MHz RISC
Fast Switching - 50-70kpps
Process Switching - 3-4kpps

And MLPP is processed switched in the 3600, just an FYI

Is the goal 3M? or is it load-balancing or redundancy, each one is different.
Todd VanDerwerken, CCNA, CCDA
Technical Consultant

http://www.ciscolinks.net

"If at first you don't succeed...then sky diving isn't for you!"

 

[multilinkguru]

hhmm, those numbers are interesting. Thanks.

For the Tasman 1400(8T1 router) or the 1200 (4T1 router), the MLPPP/MLFR processing speed is > 50Kpps with 150MHz RISC processor. And I believe they cost a fraction of the Cisco 3600 routers.

I vote against IMUX (inverse multiplex) for the simple reason that it is not as reliable or flexible as MLPPP/MLFR.

Whatever suits you right.