We have a mixture of ATM, E3 and OC3 circuits connecting to our MPLS provider. My question is regarding the QoS policy configuration applied and its calculation in relation to the bandwidth of the interfaces. For example if we have a site with 3 E1 circuits in an IMA bundle the bandwidth is 61444 Kbps but the actual CIR provided to us by the telco is 5000Kbps. Now 3 E1's right give a bandwidth of 3 x 2048 = 6144Kbps but the actual clocking on each of the ATM interfaces when I look is 1920 ; so giving 1920 x 3 = 5760Kbps
sh int atm0/0/0 | in BW
MTU 4470 bytes, sub MTU 4470, BW 1920 Kbit/sec, DLY 20000 usec
Correct me if I am wrong but the QoS policy will calcultate depending on what the Bandwidth statement on the interface is and if I dont apply one it will be the 1920Kbps ( If I didnt manually put 6144 below and left this statment out will the ATM0/IMA0 use a bandwidth of 5760 ?). Giving a wrong statement could mean reserving more bandwidth than you actually have. ( If I have 6144 configured but we have a CIR of 5000 ?). I would presume you set the BW to what the CIR is from the provider. But these circuits 'are also burstable' up to the full speed if needed so which do you use ?
What I'd like to know is which is correct and can someone explain the above to me please.
Thanks in advance for your feedback