05-08-2003 06:37 AM - edited 03-02-2019 07:12 AM
I was wondering what the rationale was, behind the "Available Bandwidth" reading when looking at the 'show int s0/0' command. I have DS1 point-to-point ckts hooked up to these three routers, with all 24 channels in use, no MUX or IMUX at all. ESF, B8ZS, all identical settings on the WICs:
C2610 running 12.2 (12a) Available bandwidth = 1158Kbit
C2611XM running 12.2 (12a) Available bandwidth = 1152Kbit
C3640 running 12.2 (7b) Available bandwidth = 1536Kbit
__________________________________________________________
How are these values calculated? I'm I not fully optimized to take advantage of my full available bandwidth?
Thanks,
Alex
05-08-2003 12:55 PM
This value is the amount of remaining bandwidth that may be set aside as "reserved" on an interface, by default, 75% of the configured interface bandwidth. Reservations may be made dynamically with RSVP, or statically using Class-Based Weighted Fair Queueing.
You can change the percentage of bandwidth that may be reserved on a certain interface using the "max-reserved-bandwidth" command in interface configuration mode. Unless you have heavy RSVP-reserved traffic or have some very specific CBWFQ needs, it is normally not necessary to adjust this value. If you are not using RSVP or other methods of reserving bandwidth, this statistic has no bearing on the function or speed of your network.
05-08-2003 01:12 PM
The parameter is informational only. The cisco device does not pick up the actual bandwidth of the link, it defaults to a value, depending on the interface type (e.g 2mbps on a WIC-1T). Set the bandwidth using the bandwidth command at interface mode, then the show interface will display the right bandwidth, and then use the 75% calculation for the reserved bandwidth.
05-09-2003 04:55 AM
Does this at all mean that my weighted-fair-queue settings are out of whack? I tweaked them to minimize input and output drops on the interface; by standard, I use:
"fair-queue 300 512 18"
PS.. I'm wondering what algorythm is used to calculate that "available for reservation" bandwidth. Why might it be different from router-to-router?
--alex
05-09-2003 05:11 AM
You're only configuring queue depth on plain WFQ (Weighted Fair Queueing) there; you're fine. It makes no bandwidth reservations, and the Available Bandwidth figure has no bearing on your configuration. All traffic has access to the entire bandwidth of the T1 subject to WFQ controlling competition with other traffic. Try this document for an explanation of CBWFQ:
Available bandwidth as shown on the interface is by default, 75% of the _configured_ interface bandwidth (i.e., interface Serial0, bandwidth 1536). This value doesn't change the actual speed of the interface, but tells the router what to expect it to be for purposes of EIGRP/OSPF routing calculations, RSVP reservations, CBWFQ configuration, etc. If you don't set it by hand, it will default to some value, which differs between various hardware types.
Examples from your first post:
1158kbit == 1544kbit * 75%
1152kbit == 1536kbit * 75%
1536kbit == 2048kbit * 75%
Some serial hardware defaults to 2MB/s configured bandwidth. 1536kbit is the "right" value for bandwidth on a point to point T1 link (64kbit * 24 channels). 1544kbit is sometimes used - this is 64kbit * 24 channels plus the 8kbit Facility Data Link channel used for out of band communication about the status of the physical T1. That channel isn't available for actual data communication, so it's most correct to use 1536kbit.
Best practice in your situation would be to set "bandwidth 1536" on each interface you have a full T1 link on, but unless you have redundant links or plan to use CBWFQ, the change will have no operational impact.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide