04-04-2002 11:59 AM - edited 03-01-2019 09:10 PM
Please help me!
I have looked over & over that the formula to configure CAR & it still makes no sense.
Can someone please explain this too me?
normal burst = configured rate * (1 byte)/(8 bits) * 1.5 seconds
extended burst = 2 * normal burst
Is the configured rate the first line of the rate limit command? IE: rate-limit 5000000
Is the configured rate applied to the formula in bits per second?
The way I seen it was.
I have a 10MB link & want to start my rate at 5MB.
So I take....
5000000 Bits per second * 1.5 seconds = 7500000 Bits per second as the committed burst /8 gives me 937500 Bytes per second which is the way you would enter it into the rate-limit command.
937500 Bytes per second * 2 = 1875000 Bytes per second for the excess burst rate.
Seems to me that now if you take all the numbers & add them together that the number is outragously higher than the orginal 10MB pipe size.
Am I nuts, or what?????
Thanks a bundle,
Dan
04-04-2002 12:27 PM
Alright....you want to rate limit an interface to 5Mbit? I'm going to ignore the formulas that you listed. There are 3 values with which we are concerned.
Avg Rate (CIR) - This is what traffic rate you want to allow through as 'conformant'. This value is in bits per second.
Burst Rate (Bc) - This value is a number of bytes. It is the Burst that is permitted and still remains 'conformant'.
Excess Burst (Be) - This value is also in bytes. This is additional burst which is the hard limit on the burst, the probability that a packet will be marked 'exceed' increases as the traffic rate approaches the Be burst value.
rate-limit
Because IP traffic is bursty, you will typically want to allow a small amount of burst to pass through and be considered as 'conforming'. This is what the Bc value is. The Be allows for traffic above the Bc, with an increasing probability of it being marked excess as it approaches Be. Same general principle as frame relay.
Hope this helps...
04-05-2002 01:29 AM
Micah ... that is a great explanation. I have a question. I have frame access @ 192k with 2 x pvc's going into it, one at 32k and one 92k. I have implemented traffic shaping so that these pvc's can burst through the CIR. If I implement the ratelimit command on the 32k pvc like for example
rate-limit 48000 4000 4000 conform transmit exceed drop
when the data bursts through the 48000 when towards the be, will it be dropped if the F/R access is being saturated or will it look at the pvc consumption. I'm sorry but I dont quite understand. Basically, on the larger pvc I have important traffic that bursts through to the access rate most of the time but some times it is quiter and doent need the bw and this is when I was hopeing to use the rate-limit command on the other pvc to limit usage on that pvc at busy times of the day or when the other pvc is busy. When the larger pvc was quiet, it could burst through to the max. Does this make sense. Could you please advise here and is there a better way of doing this?
Regards
04-05-2002 06:33 PM
You may go ahead and configure class-based policing instead of old "CAR"
http://www.cisco.com/warp/customer/105/cbpcar.html
selecting burst size
http://www.cisco.com/warp/customer/105/carburstvalues.html
also URL has plenty of links to help you configure it
04-08-2002 05:25 AM
Thanks for all the advice, but I do understand how it works.
What I want to know is what to make of the formula that I posted.
This is Ciscos recommendation & is right on the doc CD.
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