frame relay terms

Unanswered Question
Jul 2nd, 2007

hi all,

I would like to know

1. the difference between access rate, CIR, MinCir.

2. what is Tc (committed rate measurement interval), and what it has to do with burst.

correct me if I'm wrong,

access rate: is the max bandwidth at which frame relay interface can tranmit.

cir: is the guaranted bandwidth for the user.

I request someone to post a network scenario/diagram on traffic shaping.

thanking you,

prashanth

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 5 (2 ratings)
Loading.
mheusing Mon, 07/02/2007 - 05:11

Hi,

Yes, you are correct about the rates you mention above. There are a couple of rates which can be used with frame relay, which indeed can get confusing. Actually even more, because not everyone always uses the same terms. But now step by step, first about rates involved:

A) There is the rate of the access line (OSI layer 1) usually called "access rate" or "line rate"

B) The commited rate at which user traffic is policed by the service provider: "commited information rate" (CIR). In Frame-Relay those frames are usually marked with Discard Elegible bit: DE=0. In addition this is the rate specified in contracts and therein guaranteed to the customer. Note: this by no means has to be identical to internally reserved bandwidth in the FR network.

C) The excess rate allowed by some policing implementations: "peak information rate" (PIR). In Frame-Relay those frames are usually marked with Discard Elegible bit: DE=1

D) FR traffic shaping or congestion avoidance. The FR network might "ask" through BECN frames that the senders lower their rate. The router can be configured to do this, but never below minCIR.

E) FR network internal rates. In reality the SP internally might even use other rates than the ones mentioned above. He could f.e. opt to overbook the network (common practice) and reserve internally less than CIR. Those rates have different names depending on FR equipment/switches. There might be even more rates with SP internal traffic shaping and congestion avoidance.

To learn more about rates associated with policers read f.e. RFC 2697 and RFC 2698.

For a flow chart on traffic shaping, have a look at

http://www.cisco.com/en/US/tech/tk713/tk237/technologies_tech_note09186a0080093c07.shtml

Hope this helps!

Regards, Martin

bethamprashanth Fri, 07/06/2007 - 06:54

hi Martin,

firstly i thank you for responding my query:

but i need difference between cir and Mincir in simple words for me to understand so that i can go forward topics in learning traffic shaping.

i would also like to know about Tc. so that i can clearly understand Bc and Be.

Edison Ortiz Fri, 07/06/2007 - 07:06

CIR is the maximum bandwidth allowed by a frame-relay provider during times of non-congestion.

MINCIR is the minimum bandwidth allowed during time of congestion.

For instance, you pay for a frame-relay circuit and the contract states that you have a 512kbps circuit (CIR) but during time of congestion you are guaranteed 256kbps (MINCIR).

_____

TC is the time interval that packets are sent. TC value is in milliseconds and can range from 10ms to 125ms.

When using 125ms, it means that a packet is sent every 1/8 of a second. If you want to send the packets faster you change the TC value by manipulating the Bc and Be values.

bethamprashanth Tue, 07/10/2007 - 07:23

Hi All,

Thanks for valuable information u have provided. I will continue my futher study on framerelay.

also i will be greatful if someone will send me a network scenario that contain entire framerelay configuration.

scottmac Tue, 07/10/2007 - 19:23

Perhaps in the context of Cisco all the above is true.

As generic industry terms, the definitions are a little different.

Access Rate (sometimes called Committed Access Rate - CAR) is the actual MAX possible bandwidth of the link. For 512K, you are using 12 timeslots of a T1, you cannot use more than 512K, because you do not have access to more than the 12 timeslots provisioned.

Committed Informastion Rate - CIR - This is the amount of guaranteed bandwidth for your circuit / PVC. This is the amount you pay for, and the anount the provider / carrier is (usually) contractually bound to provide as minimum under any circumstance.

MinCIR is a proprietary value, there is no generic industrial parameter for MinCIR.

In the Cisco environment, this is usually the actual CIR of the PVC.

There is also a value for Comitted Burst (Bc) - the amount of burst above CIR tolerated for a given Tc. This is usually the threshold where Discard Eligible (DE)is set. Depending on the carrier, your traffic will either be discarded (regardless of congestion), maybe discarded (only if congestion is encountered), or DE is completely ignored.

Then there is Excess Burst (Be) - this is the area above Bc, and (for many/most/all carriers) the point at which your traffic will be discarded.

The basic timing interval is one (rolling)second. Tc is how that one rolling second is divided.

For slower rates, it is usually 125ms (.125 sec, 1/8th of a second). The threshold tolerance (the line of Bc/Be) for the current Tc is determined by what happened the last seven Tc; **Generally speaking** if the traffic was at or below CIR, then higher amplitude bursts are permitted and / or permitted to burst longer.

If the previous seven Tcs were above CIR, then the permitted burst (the level at which Bc becomes Be) will be lower / shorter.

For faster links (>T1/E1), Tc is frequently 10ms. The function is the same, it just evaluates a smaller time segment for the leaky bucket calculation ("Leaky Bucket" sounds stupid, but is / can be a fairly sophisticated algorithm ... try a Google on "Leaky Bucket" )

S0, if you were paying for a 512K Access Rate / CAR, with a 256K CIR, you'd have (as a starting point) an additional 256K of Bc for one Tc (Burst to 512K for Tc). Since you've "used all your burst for this (rolling) second," Be is now at CIR for the next seven Tc segments. Once that rolling second has passed, the threshold is reset and bursting (Bc for Tc) is permitted again.

The smaller / shorter the burst per Tc segment, the more burst segments, and higher amplitude burst (up to Be, with an absolute limit of CAR) is permitted.

Please note, this is a greatly oversimplified representation of a very complex topic; hopefully you get the idea.

Also note that the definitions can also vary some by Vendor (Lucent versus Alcatel versus Cisco versus ...), and the calculations for Bc, Be, and Tc and "leaky bucket" algorithms can also vary. That is one occasional source of "translation" errors at a handoff from one provider (using vendor X) to another Vendor (using Vendor Y).

Hope this helps.

Good Luck

Scott

Actions

This Discussion