I am trying to clear up some confusion. I have
a VWIC-2MFT-T1 card in my 3660 router. I have a
frame circuit in one of the interfaces. The plan
is to have more than 1 connection through this
physical interface. I know that I can do this through
the use of subinterfaces. I can configure a
subinterface for each point-to-point connection.
My confusion is about how the subinterfaces know which channels of the T1 to use. I have the controller configured to use all 24 timeslots. If,
for example, I have one connection using 56k and another using 256k. How do the subinterfaces know which channel the traffic is coming in on? Do you
configure the bandwidth statement on the subinterface
in the order that the circuits come in? For example,
do I configure subinterface 1 with a bandwidth statement of 56k and it reserves channel 1 of the
T1 for its use and then subinterface 2 has a 256k bandwitdh statement reserving the next 4 timeslots
for its use? Any enlightenment is appreciated.
Your provider will let you know what DSO to expect the circuit. For example if you have a circuit coming from remote B on DSO 7 and remote C has a 128k on channels 1 and 2
channel-group 0 timeslot 7 speed 64
channel-group 1 timeslots 1-2 speed 64
The above is the command you would configure in addition to the linecode and framing. One you do that a subinterface is automatically (and is associated by the channel group number) created by the router where you can configure an ip address etc.
Let me know if this answer your questions.
This is correct. Except that the VWIC I am using
does not support multiple channel-groups in the
NM-2W that I am using. So I have to config the
controller with all timeslots 1-24 and config
subinterfaces on that channel group. Like this,
Controller T1 2/2
channel-group 0 timeslots 1-24 speed 64
If it could support the different channel groups, I
would understand which channels are being used. Un-
fortunately it doesn't and I don't understand how
the subinterfaces know which channels to use since
the controller is set for all channels. Thanks for
the help though.
I suppose in this case you might need to look into a different hardware solution. Looking at the docs:
there is no way to use this hardware combination to support multiple fractional T1 groups. But is it what you actually need ? Since this is a Frame-Relay connection, are you sure you cannot take the full/partial T1 from the carrier and have them map the DLCI's to it ?
I am thinking I can take the partial T1s and get
DLCIs for them. I am just trying to understand how
it works. If I have a controller that has all the
timeslots allocated. It creates a Serial interface
for that channel-group. I create subinterfaces off
of that serial interface. IPs and DLCIs are assigned
to them. I believe it should work. My confusion is
how the subinterfaces know which channels of the
serial interface to use for each different frame link. Is the bandwidth statement under each subinterface the determiner of how many channels to
use? Does it matter? Because the DLCIs are configured, and the whole pipe is being used, would
the data just get to the right interface because of
the frame mapping?
Yes, you need to understand the difference between physical and logical arrangment here. Physically, you will have a single pipe, all traffic will go thru it, utilizing all the DS0's. Logically, however, traffic will be split based on the DLCI associations. You might need to look into Frame-Relay traffic shaping. As you can imagine, the central site router will be sending at a rate that is probably much higher than the remote site is setup for, up to all the DS0's aggregated rate.
So, as I understand it. The DLCIs will route the
traffic correctly and the specific channels the
traffic uses doesn't really matter. Are bandwidth
statements necessary under the subinterfaces? Is
traffic shaping necessary for this to work or is
it just an added perk? Thanks for the help.
The bandwidth statements are not required to bring up a connection. The bandwidth statement does not modify nor have any interaction with the physical speed of a particular circuit, it is only used for informational purposes. You can think of the bandwidth command as a label or a description field. The bandwidth command is only used by certain Routing Protocols (like EIGRP) to help determine which path may be preferrable. You don't even have to have the bandwidth statement match the actual line speed of the circuit. The argument that one uses in the bandwidth command is completely arbitrary.
If you had two interfaces, one a T-3 circuit and the other a T-1 circuit you could use the bandwidth statement to have a routing protocol prefer to use the T-3 circuit. If you entered a bandwidth 100 command on the interface with the T-3 and a bandwidth 10 statement on the interface with the T-1 circuit, your routing protocol would prefer to use the T-3 circuit (assuming you are using EIGRP and all other things are equal). Both circuits will still be running at their respective line speeds.
Traffic shaping would also not be required for you to establish basic connectivity. You could bring up these circuits and pass basic traffic without using traffic shaping. It would be strongly recommended that you implement traffic shaping especially if you are planning to run any kind of heavy or consistent traffic on these circuits. It's probably more than just a nice extra benefit, but you don't absolutely have to have it in order to get your WAN connections established.
I hope this answers your questions.